Please justify/explain all edits to this page, in your "edit summary" text.
<sandro> PRESENT: Ian, Boris, Smith, Zhe, PFPS, Jie, Ivan, mschnei, markus, sandro, alanruttenberg <sandro> REMOTE: rees, evan, achille, uli, bijan, christine, deborah <jar> hello <jar> I can call in for the imports discussion - if that is appropriate <jar> Just sent email to the public-wg-comments list on the subject. <jar> The meeting is starting in 1 minute, right? Am I on the right IRC channel? <jar> 0 minutes? <bmotik> Zakim, this will be owl <Zakim> ok, bmotik; I see SW_OWL(F2F)8:00AM scheduled to start 62 minutes ago <jar> hello? am I on the right irc channel? <jar> (for owlwg f2f) <Zakim> SW_OWL(F2F)8:00AM has now started <Zakim> +jar <Zakim> -jar <Zakim> SW_OWL(F2F)8:00AM has ended <Zakim> Attendees were jar <Zakim> SW_OWL(F2F)8:00AM has now started <Zakim> +jar <alanr> Eta 5 min <jar> ok, so it's IRC owl, but zakim owlwg. <jar> I'm the only one on the call now - could someone dial in from the polycom please? <alanr> Anyone elseon yet ? <pfps> scribenick: pfps <IanH> q? <IanH> zakim, who is here? <Zakim> On the phone I see jar <Zakim> On IRC I see jar, alanr, msmith, MarkusK_, Zakim, bmotik, pfps, IanH, bijan, uli, sandro, ewallace, trackbot <Zakim> +MIT346 <bijan> I'll primarily be ircing, but can, mostly, call in at key points at least until lunch <pfps> Topic: OWL RL <pfps> ian: comment 61 and 15 <pfps> ian: approve response for 15? <pfps> Proposed: send response for 15 <IanH> +1 <pfps> pfps: +1 ALU <MarkusK_> +1 <bijan> +1 <bmotik> +1 <zwu2> +1 ORACLE <msmith> +1 <pfps> RESOLVED: send response for 15 <schneid> +1 (15) <pfps> ian: lc comment 61 - why only RDF semantics for RL? <pfps> ivan: all profiles can be interpreted both ways <pfps> ian: yes, but conformance may be a bit off <pfps> msmith: conformance needs to be changed <pfps> ian: no disagreement - documents need to be clarified <pfps> boris: actual solution - what does an OWL 2 RL checker do <pfps> msmith: EL and QL are DL semantics - RL and Full are RDF semantics <pfps> ian: so we parametrize on the semantics <pfps> ian: we say that all conformance can use either semantics <pfps> ian: RL needs a bit of work for RDF syntax RL entailment checkers <pfps> schneid: RDF semantics doesn't give any benefit for EL and QL <pfps> markus: can you notice a difference in EL and QL <pfps> msmith: yes <pfps> ian: no computational guarantees except for direct semantics <pfps> msmith: we define OWL 2 RL ontology document but then don't use it <pfps> ian: need editorial fixup in conformance document (only) <pfps> ian: respond to Jos that he is right and we are fixing it in conformance <pfps> action: ian to make fixes for 61 and craft response <trackbot> Created ACTION-298 - Make fixes for 61 and craft response [on Ian Horrocks - due 2009-03-03]. <pfps> ian: conformance is parametrized on semantics <pfps> topic: OWL QL <sandro> subtopic: Add reflexive, irreflexive, and asymmetric? <pfps> ian: uli sent a message on how the submitters of the comments feel about changes <pfps> scribenick: jie <pfps> ian: profiles needs to be clear that results are for direct semantics only <MarkusK_> markus: when allowing RDF semantics for profiles, we need to make sure that the Profiles document takes this into account <MarkusK_> markus: especially the complexity part does not seem to mention this now <sandro> zakim, who is here? <Zakim> On the phone I see jar, MIT346 <Zakim> On IRC I see Jie, sandro, alanr, schneid, ivan, zwu2, jar, msmith, MarkusK_, Zakim, bmotik, pfps, IanH, bijan, uli, ewallace, trackbot <Jie> Ian: Moving again to QL <Jie> Ian: Uli is talking with the DL-Lite folks <Jie> ... and people with different and incompatible extensions <Jie> ... people agree we can't have reflexive, irreflexive and assymmetric all together <Jie> ... and we could come up with reasonably convincing use cases that why we need, as much as we can for the language as a whole, so, if it is useful, OWL in general. But we don't probably have, in the document, at this moment, any terribly convincing use case. Although, we did have some comments from some of the LC comments, said "OWL is really useful, I like to use X" as X actually points to these properties in particular. <Zakim> +Evan_Wallace <Jie> Mike: we have implementation showing these three are relatively trivial things <Zakim> +[IBM] <Achille> zakim, ibm is me <Zakim> +Achille; got it <sandro> rrsagent, pointer? <RRSAgent> logging to http://www.w3.org/2009/02/24-owl-irc 14:32:26 <RRSAgent> RRSAgent has joined #owl 14:32:27 <RRSAgent> logging to http://www.w3.org/2009/02/24-owl-irc 14:32:58 <Jie> Ian: Healthcare and Life Science WG, and Semantic Web Deployment WG both commented on usefulness of reflexive, irreflexive, assymmetric and disjoint properties. 14:33:48 <Zakim> +??P15 14:33:49 <Jie> Ivan: that is a different argument. It is one thing that these features are in OWL 2 -- i don' think there is any discussion on that -- the other things is that they certainly haven't commented to have them in QL. 14:33:49 <Jie> Ian: but they are useful. 14:33:49 <Jie> Ivan: the problem is even if it is technically possible to add such features, I'm not sure we should 14:33:57 <uli> zakim, ??P15 is me 14:33:57 <Zakim> +uli; got it 14:34:09 <uli> zakim, mute me 14:34:09 <Zakim> uli should now be muted 14:34:11 <Jie> ... add a new feature to a profile without major justification may not be good 14:34:18 <IanH> zakim, who is here? 14:34:18 <Zakim> On the phone I see jar, MIT346, Evan_Wallace, Achille, uli (muted) 14:34:19 <Zakim> On IRC I see RRSAgent, Achille, Jie, sandro, alanr, schneid, ivan, zwu2, jar, msmith, MarkusK_, Zakim, bmotik, pfps, IanH, bijan, uli, ewallace, trackbot 14:34:31 <Jie> ... I'm worry about doing that. 14:34:31 <Jie> Ian: on the plus side, none of these people care about QL, so they won't care about features in it. 14:34:44 <IanH> q? 14:34:56 <Jie> Ivan: I'm not clear, we get overall negative feedback on Profile all together. 14:36:05 <Jie> Mike: The implementation shows the three properties are easy to implement, and they allow more ontologies to be expressible in QL, which makes QL tools more useful to the community. I think we should push the language as much we can if it does not comprise the tractability -- and in this case it doesn't -- for me, I think a justification is that it is in OWL and we can add it without technical difficulty. 14:36:05 <Jie> Ian: can you get use cases from your customers? 14:36:23 <Jie> Mike: for QL, linked open data people would need it 14:36:43 <Jie> ... even though I don't have a specific example now 14:37:06 <Jie> Alan: Comments on RL says there is additional complexity of the language to learn because of the profile. Profile document needs to be more user understandable 14:37:31 <Jie> ... adding a feature to QL will not be going to impact that, and it is a new profile, we are offering a new functionality, we want to offer the best we can without adding extra trouble. 14:37:42 <sandro> zakim, who is on the call? 14:37:42 <Zakim> On the phone I see jar, MIT346, Evan_Wallace, Achille, uli (muted) 14:38:00 <Jie> Zhe: I have a question, if QL implementers implement what is described in the document, plus some more, will that make the implementation not conformant? 14:38:00 <Jie> Ian: I believe the implementation is conformant. (Peter: add more?) Yeah, I can see Peter thinks it is not, because it will find entailment that is not justified by the document. If they use a document that contains things not in the QL, then it is outside the profile. The document is not the QL document. It doesn't matter. 14:38:23 <Jie> Zhe: we can define the core stuff, and vendors can add more they need 14:38:28 <sandro> jar, ewallace, uli, Achille --- the webcam is active again -- http://www.w3.org/People/Sandro/webcam 14:38:41 <uli> +1 to Markus 14:38:55 <uli> and Mike, sorry! 14:39:26 <Jie> Markus: I agree with Mike, not because I have more use cases, but I think it is a general rule, to other 2 profiles as well, we add as many OWL 2 features as possible without compromising good computational property. It will be strange engineering if we do that for the two profiles but not to the last one -- and we know it could be done easily. In general I think it is good because, profile is aimed at making more ontology processing easier, not for having more implementation conformant. Out target should be adding more ontologies to the profile, not to having more implementation in supporting them. 14:40:03 <Jie> Ivan: Answer to Zhe, the problem is interoperability. If I write down an ontology for a given profile, I want to be sure I can run on other engines. Of course you can implement more, but it should be such that I can rely on any implementation that is conformant. 14:40:03 <Jie> ... The other thing. If we don't know what community really uses this, and we do it because it can be done, then this is not right for standardization. 14:40:48 <Jie> ... This is the argument, and that is what we have to answer to. We sure can implement more, but that's not the point. 14:41:00 <Jie> Ian: I understand that, but I don't see how it applies to the profiles in particular. We have those features, it is not the argument that these features should be in the language. If those features are useful in general, we have a reason to have them in the language, then if they can be accommodated within this profile with good computational property, then surely they should be. If they are not justified in the language, then it is another story. 14:41:57 <Jie> Ivan: The target audience for profiles is different. QL and RL are "entry-level" things in OWL 14:42:01 <Jie> ... they should be easy. 14:42:19 <Jie> Mike: One of the audiences of QL is linked data 14:42:47 <Jie> ... also include people who want to access relational data in DL structure, for instance in distributed query system. 14:42:56 <Jie> ... they are not new to OWL. So at least some targeted audience of QL wants as much as they can get from OWL. 14:43:22 <Jie> Alan: QL and RL are different. 14:43:49 <Jie> ... QL is targeted at low complexity. The more expressivity we can have in QL the better, because it will allow more to be modeled. 14:44:07 <Jie> ... The second comment is, in terms introducing people to the language, we are not designing the language to be introductory, we are designing a language suitable to the community to make them introductory, and purpose of the document is to make the profiles more accessible. 14:44:30 <Jie> Markus: I agree with Ivan we have to take into account the community we target at. If we target linked data with QL, they are very well motivated. 14:44:45 <Jie> Ivan: we are not talking about sameAs, we are talking about reflexive, irreflexive and assymmetric properties. 14:44:54 <Jie> Boris: we have some general misunderstanding why we come to profile. 14:45:05 <Jie> ... My understanding is that the profiles are technologically driven, which means you set yourself a goal, 14:45:36 <Jie> ... such as on processing database data, or low complexity or rule reasoning. Those are objective goals, then the natural thing is to have the most expressive languages to meet these goals. I think it's a misunderstanding to have profile simplified for users. I believe it is a design issue. There are different ways to present simpler versions of OWL, but that does not need to be the profile. If you take the technology driven view, these things should be here. It is not user demand driven, it is technology thing. 14:46:25 <uli> +1 to Boris -- this distinction makes "space" for vendor's PR and for teaching/KT 14:46:25 <Jie> Sandro: it reminds me of OWL Lite, which is designed to be easier to use. 14:46:34 <ewallace> This is the same discussion around OWL Lite 14:46:39 <ewallace> as Sandro says 14:46:59 <Jie> Zhe: from vendor point of view. If the WG has a specification, no matter how complex it is 14:47:08 <Jie> ... vendors are like to add more. Interoperability does truly matter. Profile, not just QL or RL, should be lean, just the core stuff that is easy. 14:48:12 <Jie> ... I don't think it is necessary to add these features. 14:49:16 <Jie> Peter: Profiles are technical things. If people don't understand, they should look at primer, not profile. 14:49:35 <Jie> Ivan: Profile is not technology driven 14:49:43 <Jie> ... it is user community and requirement driven. 14:49:54 <schneid> q+ 14:50:07 <Jie> ... it should not be the case that we do it because it is doable. I agree with Zhe to keep it as simple as possible. It does not contradict to what Peter said, profile remains a technical document, there has to be primer etc. But adding features just because it can be done is not a good argument for it. 14:51:00 <Jie> Ian: But we are throwing out features. The design is suppose to have all the features that can fit into the computational constraints. We will make a mistake not having them in. 14:51:17 <Jie> Ivan: in QL, the major disagreement is on sameAs 14:52:00 <Jie> Markus: we don't get simpler to move a feature from non-allowed list to the allowed list, especially when the features are supported by 2 of the 3 profiles anyway. 14:52:13 <IanH> q? 14:52:18 <IanH> ack schneid 14:53:06 <Jie> schneid: requirement for QL for processing database data IS a technical requirement. It is not for education of OWL -- that should be from a text book starting with some "smaller' OWL. 14:53:36 <Jie> ... features we should avoid are 14:53:57 <Jie> ... the ones need further understanding of other additional features 14:54:08 <Jie> ... or the ones may be misleading 14:54:21 <Jie> ... I think that's not the case here 14:54:23 <schneid> was owl lite easy to understand for new users? 14:54:40 <Jie> Alan: I don't think profile should be a strictly technically driven presentation. It should be more accessible. We can do it with good editors. Reducing the language does not help anybody. We should focus on how to make the document understandable. 14:55:14 <Jie> Boris: I didn't say the document should be dry, I said the design of the profile should be driven by technology. Obviously the document should be readable. 14:55:25 <IanH> q? 14:55:53 <Jie> ... To answer Ivan, I agree things should be simple. 14:56:10 <Jie> ... but it may exclude some people who wants certain things. The community is so varied, I don't think we have a good definition of what is simple. That's why we have objective criteria: technology. I do agree it is a different issue, if, for example, a particular feature is absolutely necessary, then we should say with respect to these criteria, we can't achieve that. But we should be guided by technology criteria. 14:57:00 <Jie> Ian: I agree with that. Profiles are restrictions from everything in OWL, but we need to restrict iy as little as possible. 14:57:41 <Jie> Ian: In answer to Alan, of course the document should be improved a little 14:57:49 <Jie> ... but this document is not user facing, which should be a guide. 14:58:20 <Jie> ... the specification documents are for people who build systems 14:59:21 <Jie> ... if they run into things they can't understand, there is primer, overview and (there will be) other books 15:00:00 <Jie> Alan: The problem is to reduce redundancy across documents. The syntax is a good example that tries to consolidate more than one point of view, tries to be more accessible by inclusion of examples. We should do so in profile as well. 15:00:39 <Jie> Ian: User focusing documents tell uses which profile to use, not the technical spec document like Profile. 15:01:00 <Jie> Alan: there should be at least a compact but understandable introduction. 15:01:20 <Jie> Ian: I agree. But there is limit. This document is mainly intended for implementers. We should void the mistake that pointing people to the wrong document. 15:01:40 <Jie> Alan: I think the stronger we put on that line, the harder it will be to meaningfully address people's concern about profile. Let's think about how to make the document more accessible. There should be a section on how to select a profile to use; there should be more examples, in particular negative examples: what does not fit into a profile; and some graphic things. There are something to do to make it more pleasant to approach. It will not turn it into Primer. 15:02:07 <Jie> Boris: why we need profile, it is contentious, for example QL. That's why we don't include it in the document. I really think the document is a technical definition. 15:02:50 <Jie> ... there would be extension to introduction, but it will be painful. 15:03:23 <Jie> Zhe: I agree with Alan that a good introduction will be useful. We need better explanation on choice of profiles. We should focus on accessibility. 15:03:56 <Jie> ... for a regular users without good OWL knowledge, which one to choose? 15:04:04 <Jie> ... we should make it clear to them. 15:04:12 <sandro> zakim, who is on the call 15:04:14 <Zakim> I don't understand 'who is on the call', sandro 15:04:17 <sandro> zakim, who is on the call? 15:04:17 <Zakim> On the phone I see jar, MIT346, Evan_Wallace, Achille, uli (muted) 15:04:30 <Jie> Alan: on how to explain database profiles. We don't have to say that is for database, we can explain the consequences that taking one ontologies into one profile than the other, and extending one to get other features and what happens about complexity. We can do it in neutral way. 15:05:23 <ewallace> Ian's suggestion sounds good. 15:05:45 <Jie> Ian: profile can have some discussion and point to the user facing document, but the major explanation should be in the primer. 15:05:51 <ewallace> Explain in detail in Primer, summarize in Profile doc. 15:06:38 <Jie> Ivan: I propose that Sandro puts 2 or 3 paragraphs on high level ways of using profiles. Second, some examples in Profile will help, explaining things such as what can't be done in profiles, in the same way in the syntax document. 15:06:46 <Jie> Boris: I agree 15:07:38 <Jie> Ian: we should go back and focus on decision 15:07:55 <uli> and 15:07:55 <uli> and symmetric 15:08:45 <Jie> PROPOSED: reflexive, symetric, and assymetric should be added to the QL profile. 15:08:47 <pfps> +1 (ALU) 15:08:49 <uli> +1 15:08:49 <alanr> +1 15:08:50 <MarkusK_> +1 15:08:51 <sandro> +1 15:08:52 <Jie> +1 15:08:52 <schneid> +1 15:08:52 <IanH> +1 15:08:55 <ewallace> +0 15:09:00 <ivan> 0 15:09:01 <zwu2> -0 15:09:03 <bmotik> +1 15:09:08 <bijan> 0 15:09:18 <Achille> 0 15:09:22 <msmith> +1 15:09:47 <Jie> RESOLVED: reflexive, symetric, and assymetric should be added to the QL profile. 15:10:04 <alanr> An example that I've recently dealt with on the profiles was a snomed inspired (approximate) workaround for the lack of union in EL. There are benefits and tradeoffs and this might serve as an informative example. 15:10:29 <uli> q+ to explain 15:10:37 <uli> i don't mind 15:10:41 <ivan> ack uli 15:10:42 <uli> zakim, unmute me 15:10:42 <Zakim> uli, you wanted to explain 15:10:44 <Zakim> uli was not muted, uli 15:10:50 <IanH> ack uli <sandro> subtopic: Add sameAs 15:12:14 <alanr> q+ 15:12:42 <uli> zakim, mute me 15:12:42 <Zakim> uli should now be muted 15:12:52 <IanH> q? 15:12:55 <IanH> ack alanr 15:12:57 <Jie> Uli: I want to explain the technical consequence on adding sameAs, and to propose a compromise. QL is to use RDBMS to create your data, and query it using an ontology, such as by a query rewriter. You won't touch your data, just leave the tables as they are. If we add sameAs, that is not longer possible. You can, if you have sameAs, still use a rewriter, in case your RDBMS supports recursive query. Alternatively, you can have materialized view to have sameAs tuples in the view, but that will change the database. 15:13:02 <Jie> ... I suggest to add a note in QL about it: you can add sameAs to QL, then you have to do either one or the other: materialized view or the use of recursive query. 15:13:31 <ivan> q+ 15:13:43 <Jie> Alan: I don't think not touching the database to be a requirement for QL. (schneid, Ian: it is a core requirement). For many users, having an extra table is not an issue. I would suggest to have them in the language, and add a note saying, it requires extra table if you use it. 15:13:45 <uli> no Jie, I said that, if we don't use sameAs, we can use a standard RDBMs system without touching the data. 15:14:08 <uli> ...and if we have sameAs, we need a system.. 15:14:10 <Jie> Zhe: Uli, I have a question: in argument for sameAs, is that also applicable for transitive properties? 15:14:12 <uli> zakim, unmute me 15:14:12 <Zakim> uli should no longer be muted 15:14:19 <IanH> q? 15:14:47 <msmith> no 15:14:51 <alanr> but what about using materialization for transitive 15:15:00 <Jie> Uli: your rules only need to be recursive if you want to compute sameAs, because it is a transitive relation. 15:15:17 <alanr> q? 15:15:22 <uli> zakim, mute me 15:15:22 <Zakim> uli should now be muted 15:15:28 <Jie> Ian: the question is if we have sameAs, could we also have transitive property? 15:15:30 <Jie> Uli: I can only guess, it can be materialized in the same way. 15:15:50 <Jie> Mike: sameAs reduces to reachability in undirected graph, which is provable in LOGSPACE. Transitive property is not in LOGSPACE 15:15:59 <Jie> ... it changes the complexity. 15:16:00 <Jie> Markus: In answer to Zhe: adding an extra table is not a little requirement, such as for updating. Even it is in LOGSPACE in principle, you have to do it in a smart way for updating in LOGSPACE. So having extra table doable in database is not sufficient to have it in QL. It is also true for other profiles. For RL and EL, you can also do everything in database just adding some extra tables and recursive queries. So with recursive queries you also have higher complexity with transitivity etc. 15:16:33 <uli> +1 to Markus 15:16:40 <Jie> Alan: do you really need recursive query once you get the table set up? 15:17:00 <Jie> Markus: No, initially it is recursive to fill the table, but then you can do the updating incrementally using LOGSPACE. 15:17:01 <uli> (the point about QL is "by querying only") 15:17:00 <Jie> Zhe: recursive query is supported in modern databases anyway. 15:17:10 <Jie> Markus: Yes, but it is supported with bag semantics, not set semantics. 15:18:33 <MarkusK_> markus: replying to allen, I do not think that it is enough to state that QL can still be implemented in DBs using "some extra tables" 15:19:28 <MarkusK_> markus: maintaining these tables may not be trivial, and recursive queries and "some extra tables" are sufficient for all OWL 2 profiles, so it is not a specific feature of QL 15:20:17 <MarkusK_> markus: moreover, many RDBMS may have recursive queries that implement a bag (multiset) semantics only, so termination may not be easy to achieve when relying on these queries 15:18:06 <Jie> Boris: adding transitivity is not only about recursive query, because transitive properties could have existential qualification on them. It is not about materializing some extension first using a recursive query, then querying it, because there could be some interaction through the other axioms. 15:18:36 <Jie> ... sameAs is the only thing that can introduce equivalency. 15:18:46 <alanr> zhe asked. We're done now. 15:18:47 <Jie> ... and it can be precomputed. Transitivity is much more complicated. I don't want to go into this. 15:19:00 <Jie> Ian: Transitivity is a side issue. 15:20:57 <Jie> Ivan: I feel get reinforced to know that many modern databases do actually have recursive query. My favorite approach would be that we have sameAs in QL, but make it clear for some implementations it may lead to slower response. For many users, not having sameAs will almost make this profile not useable. 15:21:30 <Jie> Zhe: I'm not arguing for or against sameAs, I'm commenting on the comment on not touching database. It is a nice thing to have, but in practice it is not always doable. For example, you have to rebuild index for better performance. Not touching everything is not that simple. 15:21:54 <Jie> Ian: That's efficiency issue. We have QL is for accessing database that may not even in your control, but just via a query interface. It's quite common. 15:22:51 <Jie> ... if we add sameas, we lose the ability to access db just from a query interface. 15:23:31 <uli> Ivan, you *do* care as a person who wants to query data through an ontology 15:23:47 <Jie> Ivan: if I'm a user, I will not care about how things are technically done. If we have sameAs, some implementation maybe slower than others that have recursive queries, as a user I don't want to know about the details. 15:25:08 <uli> Michael, it's not only related to scalability (and the index issue mentioned by Zhe might hit you), but about "what you can/want to do" before you can start querying 15:25:27 <Jie> Schneid: QL is designed for scalability, sameas may kill it. 15:26:53 <Jie> Mike: from user perspective, if we add sameAs, user may lose some access from databases that have no recursive query. 15:27:53 <Jie> Markus: QL is not a subset of EL because of inverse and symmetric properties 15:27:05 <uli> ...and existentials in the head 15:27:05 <Jie> Boris: and existentials on the LHS, and conjunctions on the RHS. 15:28:30 <Jie> Markus: The problem is, would it be a valid statement that a user of QL who wants to have sameAs, s/he might also consider switching to EL, as long as there is no inverse and symmetric properties. 15:29:14 <Jie> Ivan: The difference between QL and EL is very small. 15:29:20 <Jie> Ian: Not really, they may be close on the set of constructs, but not on language structure. 15:29:36 <Jie> Markus: Intersection of QL and RL is small. What I suggest is, instead of adding this feature, one can also switch to EL. 15:30:19 <Jie> Zhe: people need QL, not EL because they want to access database data, how can they do it with EL? 15:30:30 <Jie> Markus: I agree. 15:30:36 <Jie> Ian: I believe we should do what Uli suggested. We need to cut the discussion. 15:31:15 <sandro> Uli, do you have a handy text for your proposal? 15:32:06 <uli> yes 15:32:11 <uli> will send in a second 15:32:49 <schneid> schneid: I want to avoid to add stuff that brings a bad dilemma to implementers: if they don't support it, then they are non-conformant, and if they implement it, then their main performance advantages will break down 15:32:54 <Jie> Mike: on conformance, if a feature is not in the language, and the tool supports it, does not make the tool not conformant. 15:33:14 <uli> We propose to not add sameAs to QL, but a paragraph to its introduction that says, roughly, "hey, if you add sameAs to QL, you can't answer queries anymore using an off-the-shelf RDBMS plus a little query rewriter *without* modifying the data...but you could still answer queries by either materializing a view for the "sameAs" closure or using an RDBMS that supports recursive queries." 15:33:30 <schneid> and this dilemma would be delegated to users, of course 15:33:34 <pfps> pfps has joined #owl 15:33:45 <uli> yes 15:33:52 <Jie> PROPOSED: not add sameAs to QL, but a paragraph to its introduction that says, roughly, "hey, if you add sameAs to QL, you can't answer queries anymore using an off-the-shelf RDBMS plus a little query rewriter *without* modifying the data...but you could still answer queries by either materializing a view for the "sameAs" closure or using an RDBMS that supports recursive queries." 15:33:55 <sandro> PROPOSED: Keep sameAs out of OWL QL, in order to keep the properties of QL, with a note about QL such Uli proposes. 15:34:20 <msmith> chair adjusting wording 15:34:36 <Jie> PROPOSED: We will add sameas to the QL profile. 15:34:44 <MarkusK_> -1 15:34:44 <pfps> -1 ALU 15:34:46 <IanH> -1 15:34:46 <bmotik> -1 15:34:47 <Jie> 0 15:34:47 <alanr> 0 15:34:48 <uli> -1 15:34:49 <Achille> -100 (IBM) 15:34:51 <schneid> -0.9 15:35:01 <IanH> q? 15:35:04 <sandro> PROPOSED: Add SameAs to OWL-QL 15:35:04 <sandro> -1 15:35:06 <ivan> ack ivan 15:35:08 <msmith> 0 15:35:14 <zwu2> 0 15:35:21 <sandro> yes, bijan, following without the audio isn't really practical. 15:35:37 <Jie> RESOLVED: We will not add sameas to the QL profile. 15:36:24 <Jie> PROPOSED: add some Uli's text to profile document 15:36:28 <pfps> +1 ALU 15:36:28 <IanH> +1 15:36:28 <Achille> +1 15:36:29 <bijan> +1 15:36:29 <sandro> +1 15:36:29 <msmith> +1 15:36:29 <MarkusK_> +1 15:36:31 <ivan> 1 15:36:32 <Jie> +1 15:36:35 <alanr> +1 15:36:35 <schneid> +1 15:36:36 <bmotik> +1 15:36:42 <ewallace> +1 15:36:51 <zwu2> +1 15:37:02 <Jie> RESOLVED: add some uli's text to profile document 15:37:13 <bijan> I wonder whether if all OWL QL implementations at CR support sameAs that that would be sufficient new information 15:37:20 <IanH> PROPOSED: reflexive, irreflexive, & asymmetric properties will be added to the QL profile 15:37:34 <pfps> +1 ALU 15:37:35 <bmotik> +1 15:37:35 <MarkusK_> +1 15:37:36 <msmith> +1 15:37:37 <ivan> 0 15:37:38 <Achille> 0 15:37:38 <schneid> +1 15:37:38 <alanr> ++1 15:37:41 <uli> +1 15:37:41 <ewallace> +1 15:37:41 <sandro> +1 15:37:42 <bijan> 0 15:37:46 <zwu2> +1 15:37:54 <sandro> (this is correcting the previous proposal) 15:37:59 <Jie> +1 15:38:03 <IanH> RESOLVED: reflexive, irreflexive, & asymmetric properties will be added to the QL profile 15:38:16 <sandro> ian: (just tidying up) 15:38:20 <Zakim> -uli 15:39:15 <Jie> break 15:39:16 <sandro> Bijan, can you dial in about 3:30 boston time? 15:39:36 <Zakim> -Evan_Wallace 15:39:37 <bijan> So 8:30 here, yes? 15:39:38 <bijan> Probably 15:46:59 <Zakim> -Achille 15:54:49 <pfps> no updated agenda - we are still finishing yesterday's agenda 15:55:28 <jar> oh my. so you will all have to extend your stay so you can finish today's tomorrow :-) 16:01:09 <sandro> scribe: Boris 16:01:36 <Zakim> +[IBM] 16:01:39 <bmotik> topic: TQ comments 16:01:54 <bmotik> ianh: I've drafted a response 16:02:01 <IanH> http://www.w3.org/2007/OWL/wiki/LC_Responses/JC1b 16:02:35 <bmotik> ianh: I've tried to tease out each of the individual comments that had technical content 16:04:48 <bmotik> (everyone's reading Ian's response) 16:07:49 <msmith> editorial comment: s/IEFT/IETF/g (Internet Engineering Task Force) 16:08:55 <bmotik> ianh: Let's go through the comment 16:09:23 <bmotik> ivan: There were specific comments by TQ that we should stop the OWL 2 effort altogether 16:09:36 <bmotik> ianh: My response does not address this 16:09:55 <bmotik> ianh: We are currently disucssing only the technical comments from Jeremy's e-mail 16:10:33 <bmotik> ianh: My response should say that there will be another response about the philosophical objections 16:10:47 <bmotik> ianh: Thanks -- I'll add this to my response 16:11:50 <bmotik> ianh: Links to Wiki's should be the links to TR 16:12:02 <bmotik> sandro: I can't find these links, but I'll ask Jeremy 16:12:27 <sandro> action: sandro find and fix the to-wiki-links Jeremy complains about 16:12:27 <trackbot> Created ACTION-299 - Find and fix the to-wiki-links Jeremy complains about [on Sandro Hawke - due 2009-03-03]. 16:12:56 <bmotik> ianh: I'll make the comment about syntax examples more precise w.r.t. what we decided at this F2F 16:15:33 <bmotik> schneid: Jeremy says that various disjointness axioms would make implementation more difficult 16:16:35 <bmotik> ianh: Rather than just making statements "It's easy to implement", can we point to implementations? 16:16:47 <bmotik> ianh: Zhe, does your implementation support disjoint union? 16:16:49 <bmotik> Zhe: No 16:17:35 <bmotik> bmotik: OWL 2 RL does not have disjoint union, but does have disjoint properties 16:18:21 <bmotik> ivan: We can just say that we don''t understand why disjoint union would be difficult to implement 16:18:30 <bmotik> ivan: We could ask for more explanation 16:19:09 <bmotik> alanr: We already said that disjoint classes have benefits, but what to say aout the disjoint union? 16:19:58 <bmotik> ianh: We'll tweak the proposal to say that this does not address all the points and say that we don't see the difficulty in implementations 16:20:37 <bmotik> ianh: In OWL 1, there was some OWL file that was used to capture bits of RDF 16:21:09 <bmotik> schneid: There is no technical need to add this: (1) no sense on the DL side and (2) it is entailed by the full side 16:21:16 <bmotik> ianh: We'll add this 16:21:39 <bmotik> ianh: Jeremy suggested changing the serialization of property chains 16:22:18 <bmotik> pfps: No, they are suggesting something else 16:22:39 <bmotik> pfps: RDF allows blank nodes in properties 16:23:01 <bijan> RDF doesn't allow blank nodes in properties. 16:23:41 <bmotik> alanr: Jeremy is worried about a blank node being used as subject or object that will then get turned into a property by some rule 16:24:21 <MarkusK_> markus: the fact that predicates in RDF cannot be bnodes is not a bug but a feature: 16:24:24 <bmotik> msmith: Jeremy doesn't point this out, but does not this also imply that bnodes are not good for inverse properties 16:24:25 <pfps> RDF does not allow bnodes for predicates - it allows bnodes for properties 16:24:44 <MarkusK_> markus: we explicitly do not want anybody to use the bnode property of some OWL 2 property chain in a triple 16:24:53 <bijan> Oh, right. Yes. Carry on. _:x rdf:type rdf:Property 16:25:06 <MarkusK_> markus: since this would be a statement about the property chain that is not supported by OWL 2 anyway 16:25:26 <bmotik> schneid: I was careful on the Full side to avoid the bnode to become a property chain 16:25:29 <MarkusK_> markus: effectively, it would be similar to allowing inverted property chain inclusions 16:25:41 <bmotik> schneid: The full semantics does not make this LHS property into a property chain 16:26:01 <bmotik> schneid: The bnode does not represent a property chain 16:26:19 <bmotik> schneid: I believe that people will be confused by this 16:27:07 <bmotik> schneid: We overloaded the rdfs:subPropertyOf to do something that it wasn't designed for 16:27:14 <bmotik> schneid: I couldn't find a real problem 16:27:46 <bmotik> schneid: I'd like to have a single triple encoding 16:28:23 <bmotik> schneid: On the LHS would be a superproperty, and on the RHS would be a list with the chain 16:28:38 <bmotik> ianh: What do we think of this? 16:28:42 <bmotik> bmotik: I don't care 16:29:02 <bmotik> ianh: Didn't we have an issue about this? 16:29:13 <bmotik> schneid: I had it on my agenda, but didn't want to bring it up 16:30:07 <bmotik> ivan: I remember that, when I needed to familiarize myself with the property chains, the current encoding was complicated 16:31:31 <bmotik> PROPOSED: Change the encoding of the property chains to a single-triple encoding (LHS is the superproperty and RHS is the list of properties) 16:31:47 <schneid> +1 16:31:50 <MarkusK_> +1 16:31:51 <bmotik> bmotik: +1 16:31:52 <pfps> -0 16:31:52 <alanr> +1 16:31:58 <ewallace> 0 16:32:08 <bijan> 0 16:33:31 <bmotik> (Addendum: it will be called owl:propertyChainAxiom) 16:34:03 <Jie> 0 16:34:04 <MarkusK_> +1 16:34:07 <alanr> +1 16:34:11 <ewallace> +1 on owl:propertyChainAxiom name 16:34:12 <schneid> +1 16:34:16 <msmith> 0 16:34:16 <Achille> 0 16:34:18 <ivan> +1 16:34:20 <IanH> 0 16:34:21 <sandro> 0 16:34:30 <zwu2> +1 16:34:32 <pfps> +0.2 for chaining the property axiom 16:34:33 <bijan> 0 16:34:34 <bmotik> RESOLVED: Change the encoding of the property chains to a single-triple encoding (LHS is the superproperty and RHS is the list of properties) -- with the addendum 16:35:23 <bmotik> schneid: Note that owl:propertyChain gets ditched 16:35:41 <bmotik> ianh: TQ complained about negative property assertions 16:35:50 <bmotik> alanr: Nobody compained about them 16:36:00 <bmotik> ianh: Some people found them useful 16:36:17 <bmotik> schneid: He had a problem with the encoding and with the negative tiples 16:36:30 <bmotik> ianh: What about my response? 16:36:40 <bmotik> alanr: I'm good with this 16:36:48 <bmotik> ianh: So that covers it? 16:37:00 <bmotik> ianh: OK, so let's move on to SelfRestrictions 16:37:44 <bmotik> alanr: Local reflexivity is more useful than the global reflexivity 16:37:59 <bmotik> schneid: In the past, there was a problem with certain semantics 16:38:33 <bmotik> schneid: Now, however, the paradox is no longer pertinent 16:38:55 <bmotik> ianh: So we can strenghten the response by saying that local reflexivity is more useful than the global one 16:39:21 <bmotik> schneid: THis is particular in RDF 16:39:32 <bmotik> ianh: And we say that there is no problem now as paradoxes do not arise 16:40:34 <bmotik> ianh: Jeremy doesn't like reflexive, irreflexive, asymmetric, and disjoint properties in general 16:41:00 <bmotik> alanr: Can't we add a line to the response saying that we'll extend NF&R? 16:41:08 <ewallace> Holger had this same position before Jeremy joined TopQuadrant 16:41:19 <bmotik> markusk: Have we got any use-cases for globally reflexive properties? 16:41:58 <bmotik> ianh: So global reflexivity approximates local reflexivity, particularly in the profiles that don't have local reflexivity 16:42:23 <alanr> this point should be added to NF&R 16:42:31 <schneid> schneid: global reflexivity can be used for local reflexivity in profiles which do not have local reflexivity (QL): e.g. to approximate locatedIn property to be "locally" reflexive" on class "Location" 16:42:56 <bmotik> ianh: We'll say that we'll clarify this in NF&R 16:43:14 <bmotik> ivan: We should add this to the introductory text 16:43:59 <bmotik> ianh: I'll say that we'll extended NF&R 16:44:01 <bmotik> ianh: Let's move to OWL/XML 16:44:26 <bmotik> ivan: When you say that OWL/XML is not a new feature -- Jeremy probably knows that it is not a new feature 16:44:39 <bmotik> ivan: Jeremy is not satisfied with the recommendation status 16:45:11 <bmotik> alanr: Can we have a small section in NF&R explaining why we want OWL/XML? 16:45:18 <bmotik> alanr: Bijan has a coherent story 16:45:39 <bmotik> ianh: Good, we'll add this and mention this addition in the response 16:46:01 <bmotik> pfps: We can say "There is rational for it and wil lbe (has been?) added" 16:46:21 <bmotik> ivan: The sentence about "not a new feature" should go 16:46:42 <bmotik> ianh: The next thing is Manchester Syntax 16:46:51 <bijan> For NF&F or whatever, here's my earlier bit: <http://www.w3.org/mid/9926856B-8AF7-4F74-89DC-6C3AEE607EC9@cs.man.ac.uk> 16:46:59 <bijan> on OWL/XML 16:47:13 <bmotik> (everyone): ship it 16:47:34 <bmotik> ianh: Jeremy doesn't like using reification in annotations 16:48:10 <bmotik> bijan: I don't recall any explicit feedback about reification 16:48:39 <bmotik> bijan: We used our own vocabulary to avoid overloading the meaning of the RDF vocabulary 16:49:21 <bmotik> ianh: Jeremy is worried about reification at all 16:49:40 <bmotik> ianh: But this doesn't handle annotation on axioms 16:49:57 <bijan> We considered *many* alternative encodings, e.g., Literals 16:50:24 <bmotik> ianh: The response says that, if a single axiom is annotated, there is nothing to hang the annotation off of 16:50:32 <bmotik> ianh: Therefore, we *must* reify 16:50:49 <bmotik> ianh: I pointed to our discussion about the usage of RDF reification 16:51:04 <bmotik> schneid: Raised by Jeremy! 16:51:22 <bmotik> ianh: So we're happy with the response as is? 16:51:30 <bmotik> alanr: I hear no objections 16:51:58 <bmotik> ianh: I could only make it clearer that we do hang annotations off of blank nodes whenever there is one 16:52:07 <bmotik> ianh: Other than that, we are good with it 16:52:28 <bmotik> ianH: Moving on to n-ary datatypes 16:52:36 <bmotik> alanr: I have a problem with how this is stated 16:53:19 <bmotik> alanr: We should say that we introduced hooks because there was a reasonably thought out extension that will be presented as a note, but not say too much what you can do with it? 16:53:35 <bmotik> ianh: Let's skip on the next one while Alan is generating text 16:53:49 <bmotik> ianh: Moving on to RDF interoperability 16:54:52 <bmotik> ivan: Looking at the comment itself, my feeling is that it falls in the same caterogy of general misunderstanding regarding the role of RDF 16:55:00 <bmotik> ivan: We have already addressed that 16:55:23 <bmotik> ivan: We should say that the overall structure has not changed a bit compared to OWL 1 16:55:31 <bmotik> ivan: I would simply say "Nothing has changed" 16:55:55 <bmotik> ianh: I can strengten the second sentence in my proposed response 16:56:09 <bmotik> ivan: I see that you are referring to some other responses 16:57:02 <bmotik> ivan: Sorry, not important 16:57:20 <bmotik> alanr: Why are we saying that the role of RDF is better than it was? 16:57:33 <bmotik> ivan: It is the same, not better, not worse 16:58:01 <bijan> Tactically, it's better not to say "better" because that gets us into a debate about whether it's *really* better 16:58:03 <bmotik> ianh: Alan is saying that we could improve interoperability (by taking up more graphs), but we don't go there 16:58:07 <bijan> "not changed" is less arguable 16:58:37 <bmotik> ianh: Appendix and dependcies on life sciences 16:58:45 <bmotik> alanr: We should response a bit more actively 16:59:14 <bmotik> alanr: We should say that we'll explore the possibilities for diversifying the examples in NF&R 16:59:21 <bijan> zakim, mute me 16:59:21 <Zakim> bijan was already muted, bijan 16:59:30 <bmotik> alanr: We should also say that we welcome examples from his user base 16:59:44 <bmotik> ianh: He complained about some trivial typos 17:00:52 <bmotik> ianh: Another complaint was that NF&R motivated features that are not in OWL 2 17:01:01 <bmotik> ianh: It is similar to OWL 1 17:01:19 <bmotik> ianh: We motivated certain features, but not included all of them 17:01:26 <bmotik> alanr: Why don't we get rid of them? 17:01:37 <bmotik> ianh: It could be useful to document them 17:01:46 <bmotik> ianh: I'd be OK with deleting these 17:02:02 <bmotik> pfps: We were supposed to gather use cases and requirements 17:02:15 <bmotik> pfps: This is what we did and should not be throwing away our work 17:02:18 <bijan> Throw it away! 17:02:27 <bmotik> alanr: The document is called "New Features and Rationale" 17:02:34 <bijan> The use cases right? I'm strongly against them 17:02:34 <bmotik> alanr: These are not new features 17:02:56 <bmotik> pfps: Given the abstract of the current document, Alan is correct 17:03:19 <bmotik> ianh: The document wasn't supposed to be a general "Use Cases and Requirements" document 17:03:49 <bmotik> PROPOSED: Remove UC10 and UC11 from NF&R 17:04:19 <bmotik> ewallace: I was just wondering we're still controversial about the n-ary hook 17:04:38 <bmotik> ewallace: This is a motivation for n-ary 17:04:47 <bmotik> ianh: This is a good point 17:05:13 <bmotik> ianh: Evan is saying that motivating the hook for n-ary is not bad 17:05:37 <bmotik> alanr: If it speaks to what we have in the n-ary note, I'm OK with that 17:05:49 <bmotik> ianh: I believe that UC10 and UC11 will be covered by the note 17:06:00 <bmotik> alanr: Then we can say that this is the motivation for the note 17:06:48 <bmotik> ianh: The response to Jeremy then becomes that these use cases motivate the hooks 17:06:57 <bmotik> alanr: I'd say that they motivate what's in the note 17:07:06 <bmotik> ianh: Alan should craft the text for that 17:07:55 <bmotik> ianh: Some references to TQ composer were fixed 17:08:15 <bmotik> ianh: Jeremy doesn't like Manchester syntax 17:08:55 <bmotik> pfps: If the WG decides that there will not be MIME type for Man syntax, it will happen anyway 17:09:08 <bmotik> bijan: I'm not sure whether one can comment on a note 17:09:57 <bijan> zakim, mute me 17:09:57 <Zakim> bijan should now be muted 17:10:08 <bmotik> bijan: We could say "This will not be a REC document. THanks for the comment, but we won't follow it" 17:10:19 <bmotik> ianh: Next is GRIDDL 17:10:36 <bijan> I didn't hear that 17:10:57 <bmotik> ianh: My response says that the charter does not mandate GRIDDL 17:11:07 <bmotik> alanr: This is not a general reading of the charter 17:11:12 <bmotik> ivan: I agree 17:11:23 <bijan> I'm happy with that response 17:11:29 <bmotik> ivan: My proposal is to say that this is still a subject of an open issue 17:11:29 <bijan> (ivan's) 17:11:40 <bijan> I'm off again 17:11:49 <bmotik> bijan: I agree with Ivan's rpoposal 17:12:20 <bmotik> ianh: OK. THe response will be "This is a subject of an open issue, and we'll take your opinion into consideration" 17:13:21 <bmotik> ianh: The next comment is again about normativeness of OWL/XML 17:14:02 <bmotik> msmith: IETF has it own notions about normative and informative and these are disconnected from MIME type registration 17:14:21 <bmotik> msmith: I'll look up a reference 17:14:34 <bijan> MIME type registration is normative *for that type*, not that the W3C has made it noramtive. N3 has a mime type! 17:14:43 <bmotik> ianh: The response to this will be to say "The XML syntax is optional" 17:15:14 <bmotik> pfps: He also appears to be complaining that the document is REC rather than a note 17:15:33 <bmotik> sandro: In my mind it is logically nonsense to have a specification which is nonnormative 17:15:46 <IanH> ack bijan 17:15:46 <pfps> normative is not the same as rec-track 17:15:55 <bmotik> bijan: Jeremy raised several points 17:16:09 <bmotik> bijan: I have plenty of motivation for XML syntax 17:16:33 <bmotik> bijan: We have also done our best not to be divisive 17:16:47 <bmotik> bijan: We are reaching to the rest of the world (such as XML) 17:17:30 <bmotik> bijan: We'd registed a MIME type even if XML syntax were a note 17:18:05 <bmotik> bijan: We should say that we want to have a single XML-friendly exchange format 17:18:39 <bmotik> ianh: Could you type into IRC some text about these points? 17:18:43 <bmotik> bijan: I'll do it 17:18:49 <bijan> I think this should be the response to JJC 17:19:03 <bmotik> ivan: There is already an entry on OWL/XML and we are repeating here a part of our reponse 17:19:10 <bijan> 1) Motivation: XML toolchain friendly owl foramt (e.g., SOAP, etc.) 17:19:16 <bmotik> ivan: I don't see a need for repetition 17:19:29 <bijan> 2) Divisive, it helps bridge the gap between the XML world and semantic web world 17:19:51 <bmotik> ivan: I think we can simply refer to the Document Overview that will describe the place of OWL/XML in the grand scheme of things 17:19:51 <bijan> 3) Why recommendation? Because we want to standardize the XML toolchain friendly owl format 17:20:02 <bijan> Fine 17:20:28 <bmotik> alanr: Less is more, Bijan. I don't agree with your particular arguments, but we don't need to include them 17:21:14 <bmotik> ianh: We'll have one oint response about XML. We've already decided on what that is. 17:21:25 <bmotik> ivan: We can only refer to the Document Overview. 17:21:27 <baojie> baojie has joined #OWL 17:21:34 <bmotik> ianh: Moving on to owl:real 17:21:52 <msmith> the relevant reference to media type registration and the relationship to normativity from IETF's perspective is http://www.ietf.org/rfc/rfc4288.txt section 4.10 17:21:56 <bmotik> ivan: We can't do anything here because it is pending resolution of issues from yesterday 17:22:16 <bmotik> ianh: We go back to the cases where Alan was asked to craft some text 17:22:30 <alanr> UC#10 and UC#11 motivate a feature which the working group was not able to fully develop, but for which we have published a note [cite note]. 17:22:36 <alanr> N-ary datatype: This specification currently does not define data ranges of arity more than one; however by allowing, syntactically, for n-ary data ranges, the syntax of OWL 2 provides a "hook" allowing the working group to introduce experimental extensions as will be published as in [cite note]. 17:24:09 <bmotik> ianh: Good, we're done with that 17:24:50 <bmotik> ianh: There were a couple of comments that were between technical and motivational. I'd like to ask for some advice on that 17:25:00 <bmotik> ianh: One comment is regarding effactiveness 17:25:14 <bijan> Isn't the abstract going to change? 17:26:48 <bmotik> ianh: Jeremy doesn't like the abstract of the document mentioning effective reasoning algorithms 17:26:58 <bmotik> ianh: The response is "We'll rewrite the abstract" 17:27:20 <bijan> I don't think we should get into a debate with him about the word "effective" 17:27:39 <bmotik> pfps: We'll remove the offending word from all documents apart from the Profiles (where it has a particular meaning) 17:27:41 <bijan> He supports OWL Full! :) 17:27:48 <bmotik> ivan: It is ducking his comments. 17:28:01 <bmotik> ivan: I don't know what to asnwer regarding his non-belief 17:28:39 <ewallace> Isn't less still more? 17:28:50 <bijan> Even less is way more 17:29:23 <bmotik> alanr: The charter doesn't talk about "effective", but "reasonable" and "feasible" 17:30:03 <ewallace> +1 17:30:11 <bmotik> ianh: Our response is "The abstract has changed, and we no longer talk about 'effective'" 17:30:15 <bijan> +1 17:30:19 <bmotik> ianh: His next comment is more philosophical 17:30:42 <bmotik> ianh: We made a lot of mention of the OWL-ED workshop and that this didn't represent a broad spectrum of the OWL community 17:31:04 <ewallace> It was in NF&R 17:31:05 <bmotik> ivan: We should not mentioned OWL-ED anywhere, and I don't think we have any mention of it in our documents 17:31:19 <bmotik> alanr: I thinnk it is appropriate to mention OWL-ED in references, but nowhere else 17:31:53 <bmotik> (everyone looking at NF&R) 17:31:58 <ewallace> It is still there. 17:32:14 <bmotik> pfps: It is in the overview but in a completely unobjetionalbe spot 17:32:21 <bmotik> pfps: We could change "much" to "some" 17:32:41 <bmotik> alanr: In the intreset of less-is-more, I don't see a problem with removing it 17:32:47 <bmotik> pfps: I think it belongs in that paragraph 17:32:52 <bmotik> sandro: I agree 17:33:06 <bmotik> bijan: It is a comment about a non-LC document and it is a non-technical comment 17:33:25 <ewallace> This one will go to Last Call. 17:33:32 <schneid> +1 to bijan (in general for non-lc docs) 17:33:41 <bmotik> bijan: We could say "Thanks for the comment, but this is a manner of editorial discression; you can comment at LC" 17:33:52 <bmotik> ivan: We are just postponing this issue. This doens't make much sense 17:34:10 <bmotik> ivan: Instead of "much" we say "some" and this seems quite good 17:34:26 <bmotik> bijan: I'd be perfectly happy for them to raise a new LC comment and to give the same response 17:35:13 <bmotik> ianh: I think everything feels that changing "much" to "some" would be sufficient 17:35:26 <bmotik> alanr: But what do we lose if we remove it? 17:35:26 <bijan> I think it's fair and helpful 17:35:44 <bmotik> pfps: We remove the connection to our history! TQ wants to revision history! 17:36:21 <bijan> I think it's a denial of service attack. I vote with the majoirty 17:36:22 <bmotik> alanr: I love OWL-ED. I just believe that the connections to the OWL-ED are reflected with references 17:36:37 <bmotik> PROPOSED: The reference to OWL-ED stays in the document but with a change of "much" to "some" 17:36:39 <bmotik> bmotik: +1 17:36:39 <pfps> +1 ALU 17:36:40 <msmith> +1 17:36:43 <ewallace> -1 17:36:45 <ivan> +1 17:36:45 <IanH> +1 17:36:47 <MarkusK_> +1 17:36:47 <alanr> -1 (but won't block) 17:36:47 <zwu21> 0 17:36:47 <schneid> +1 17:36:50 <baojie> 0 17:36:58 <sandro> +1 17:36:59 <bijan> 0 17:37:10 <Achille> 0 17:37:15 <IanH> q? 17:37:27 <bmotik> ewallace: I would go with Bijan and Peter 17:37:40 <bmotik> ewallace: I voted against changing "much" to "some" 17:37:47 <bmotik> ianh: Will you lie in the road? 17:37:50 <bmotik> ewallace: No 17:38:03 <bmotik> RESOLVED: The reference to OWL-ED stays in the document but with a change of "much" to "some" 17:38:17 <bmotik> ewallace: Ask Christine to make the change 17:39:34 <bijan> Who's changing it? 17:40:17 <bmotik> bmotik: I've changed "much" to "some" 17:41:20 <bijan> Earlier for Bijan-issues would be appreciated 17:41:25 <ewallace> +1 on replanning now 17:47:01 <ewallace> Don't worry about me, time wise. 17:47:47 <ewallace> What time are we planning for the NF&R discussion? 17:48:12 <ewallace> Just want to know when to encourage Christine to join. 17:53:40 <baojie> Ian just said "Other Documents" will be discussed 18:30:35 <bijan> christine, my (jokey) comment was directed at the use cases, not NF&R or n-ary 18:30:37 <bijan> Sorry for the confusion 18:34:21 <zwu21> scribenick: Zhe 18:34:22 <ivan> zakim, who is here? 18:34:22 <Zakim> On the phone I see MIT346 18:34:23 <Zakim> On IRC I see schneid, alanr, msmith, pfps, christine, sandro, baojie, zwu21, RRSAgent, Achille, ivan, jar, MarkusK_, Zakim, bmotik, IanH, bijan, ewallace, trackbot 18:34:26 <pfps> scribenick zwu21 18:34:45 <sandro> scribe: Zhe 18:35:12 <zwu21> ... 18:35:13 <alanr> http://www.w3.org/2007/OWL/wiki/LC_Responses/TC1 18:35:23 <zwu21> scribenick: Zhe 18:35:47 <zwu21> Topic: philosophical 18:36:11 <zwu21> alanr: goal is to look at responses that have been drafted 18:36:12 <Zakim> +??P5 18:36:17 <bijan> zakim, ??p5 18:36:17 <Zakim> I don't understand '??p5', bijan 18:36:22 <bijan> zakim, ??p5 is me 18:36:24 <Zakim> +bijan; got it 18:36:26 <bijan> zakim, mute me 18:36:26 <Zakim> bijan should now be muted 18:36:30 <zwu21> ... looking at TC1 18:36:42 <zwu21> ivan: only one change made. 18:36:58 <zwu21> ... last sentence before the refences 18:37:08 <IanH> q? 18:37:18 <Zakim> +Evan_Wallace 18:37:33 <IanH> we don't hear any noise 18:38:23 <ewallace> Hearing nothing. 18:38:45 <alanr> structural specification and functional-style syntax document 18:38:57 <zwu21> alanr: make a normal reference 18:39:14 <pfps> abstract structure changes to generic syntax 18:39:14 <alanr> drop 18:39:19 <alanr> drop "This was only a matter of timing; the plan is to have both semantics (and all other documents) published as Recommendations together." 18:39:34 <alanr> was not _yet_ published changes to has not yet been published 18:40:02 <pfps> OK by me 18:40:12 <alanr> q? 18:41:16 <zwu21> ivan: will send it out tommrrow 18:41:41 <alanr> http://www.w3.org/2007/OWL/wiki/LC_Responses/FH3 18:41:48 <zwu21> ivan: regarding LC 29, 18:42:23 <zwu21> pfps: there are two responses. we are both stuck 18:42:30 <zwu21> ... with Bijan's 18:42:48 <bijan> I give up mine without hesitation 18:43:03 <bijan> I didn't put it in there but sent it to the list 18:43:27 <bijan> Mine is more on justifying xml syntax 18:43:36 <bijan> Peter's is more about the harmlessness of owl/xml 18:44:17 <christine> if still plan to discuss Documents, at what time please ? 18:44:18 <zwu21> IanH: we agreed on a bare minimal response to TopQuadrant's comments 18:44:35 <bijan> +1 18:45:04 <pfps> OWL/XML: XML syntax is not a new feature -- see . It should also be noted that RDF/XML is the only syntax that MUST be supported by implementations; support for the XML syntax is not required (see also FH3). 18:45:10 <zwu21> pfps: ... jc1b 18:45:34 <zwu21> alanr: add a note that we will add something in NF&R 18:45:49 <sandro> amended to (1) remove the XML syntax is not a new feature, and (2) link to NF&R 18:46:16 <zwu21> ivan: the reason I think short resposne is ok 18:46:58 <bijan> Cool! 18:47:06 <bijan> Then I'm all for microshort 18:47:09 <alanr> PROPOSED: Respond to FH3 as in JC1b 18:47:29 <bijan> I'mhappy to be out of the loop here 18:47:39 <alanr> +1 18:47:43 <Zakim> +[IBM] 18:47:45 <zwu21> pfps: delegate to IanH for response 18:47:46 <bijan> +1 18:47:47 <msmith> +1 18:47:48 <MarkusK_> +1 18:47:49 <zwu21> +1 18:47:56 <Achille> zakim, ibm is me 18:47:56 <Zakim> +Achille; got it 18:47:59 <sandro> +1 18:48:13 <alanr> RESOLVED: Respond to FH3 as in JC1b 18:48:26 <pfps> +1 18:48:28 <zwu21> subtopic: LC 34A 18:48:37 <alanr> http://www.w3.org/2007/OWL/wiki/LC_Responses/JC1a 18:48:38 <baojie> +1 18:48:54 <zwu21> alanr: I hope we can have something shorter 18:49:01 <alanr> http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/att-0051/index.html 18:49:05 <zwu21> pfps: fine by me 18:49:06 <bijan> Second paragraph only? 18:49:12 <bijan> First and second paragraph only? 18:49:32 <bijan> Me 18:49:51 <zwu21> IanH: bijan wrote the initial version. 18:49:58 <zwu21> ... some of it is used here 18:51:01 <zwu21> ivan: this is the answer to his comment to stop the work? 18:51:16 <alanr> q? 18:51:25 <zwu21> ... can we add something more formal? 18:51:33 <bijan> q+ 18:51:49 <zwu21> ... for example, a few WG members want to move forward 18:51:52 <bijan> zakim, unmute me 18:51:52 <Zakim> bijan should no longer be muted 18:51:54 <alanr> ack Bijan 18:52:02 <IanH> ack bijan 18:52:12 <alanr> q+ sandro 18:52:36 <zwu21> bijan: I don't see that TopQuardrant wants us to stop work 18:52:41 <alanr> q+ ianh 18:52:48 <zwu21> ... he asked that we redo all the work we have done 18:53:00 <zwu21> ... according to the process he think is more appropriate 18:53:16 <alanr> q? 18:53:19 <zwu21> ... We can safely ignore it 18:54:09 <zwu21> ... given the strong support from lots of WG members, we can just let it go 18:54:09 <alanr> ack sandro 18:54:14 <bijan> zakim, mute me 18:54:14 <Zakim> bijan should now be muted 18:54:27 <zwu21> sandro: I am for short responses 18:54:40 <zwu21> ... not sure what we can do differently here 18:54:45 <alanr> ack ian 18:54:46 <zwu21> pfps: you can just point to NF&R 18:54:50 <alanr> q+ ivan 18:54:54 <alanr> q+ alanr 18:55:06 <zwu21> ianH: a) one of the option is to stop working on OWL and start working on something else 18:55:10 <bijan> Oh, WebSHROIQ 18:55:11 <bijan> I see 18:55:14 <zwu21> ... and don't call it OWL 18:55:37 <bijan> zakim, unmute me 18:55:37 <Zakim> bijan should no longer be muted 18:55:40 <alanr> q? 18:56:08 <alanr> ack ivan 18:56:13 <bijan> zakim, mute me 18:56:13 <Zakim> bijan should now be muted 18:56:20 <sandro> bijan: Put third paragraph of http://www.w3.org/2007/OWL/wiki/LC_Responses/JC1a into Positive Last Call Responses web page 18:56:23 <IanH> I like that -- put positive responses on a wiki page 18:56:41 <zwu21> ivan: what I would do to the last paragraph is to list the references (positive comments about the features) 18:56:45 <IanH> That way we could even be more expansive 18:56:51 <zwu21> ... and they can read/check it 18:56:55 <sandro> "blurbs" 18:56:56 <bijan> I'd like the testimonal page anyway 18:56:59 <ewallace> +1 to putting positive comments on a page and including a pointer to that 18:57:11 <sandro> +1 to a testimonial/blurbs page 18:57:18 <IanH> q+ 18:57:22 <zwu21> ... I still believe that some kind of statement says that based on these positive comments, WG should move forward 18:57:24 <sandro> (W3C usually does it during PR, but we can start now.) 18:57:33 <zwu21> ... according to the charter 18:57:36 <alanr> ack alanr 18:58:37 <alanr> ack ianh 18:58:38 <zwu21> alanr: suggest 1) chaning course is not an option; 2) point out positive comments 18:58:46 <zwu21> s/chaning/changing/ 18:58:52 <bijan> q+ 18:59:26 <alanr> q? 18:59:42 <bijan> q- 18:59:51 <alanr> q? 18:59:52 <zwu21> IanH: first thing is to align with JC1B response, we would improve the motivation. make it more constructive 19:00:05 <zwu21> ... for the rest, point to a web page 19:00:10 <IanH> q? 19:02:05 <zwu21> pfps crafted FH1 response 19:02:31 <zwu21> Topic: document schedule 19:02:34 <alanr> q? 19:02:47 <bijan> zakim, mute me 19:02:47 <Zakim> bijan was already muted, bijan 19:02:52 <zwu21> ivan: what I believe is in the next roudn of publications, we 19:03:01 <zwu21> ... do a complete publication of all our documents, 19:03:25 <zwu21> ... the current LC documents to be re-issued as LC 19:03:25 <alanr> q? 19:03:50 <zwu21> ... for the current working drafts, we should republish them as working drafts 19:04:04 <zwu21> ... hope that RDF semantics could be LC, quick reference be LC 19:04:25 <zwu21> ... ok with re-issue another draft of Primer 19:04:35 <zwu21> ... not sure about NF&R, 19:04:45 <bijan> q+ 19:05:09 <zwu21> ... politically, re-issue everything as a package, without implying a priority, is the right thing to do 19:05:18 <alanr> q+ ianh 19:05:30 <zwu21> ... regarding timing, RDf semantics is not clear to me 19:05:50 <Zakim> +??P8 19:05:51 <zwu21> Michael: when do you think is the earliest date for publishing? 19:05:57 <IanH> q? 19:06:06 <IanH> zakim, who is here? 19:06:06 <Zakim> On the phone I see MIT346, bijan (muted), Evan_Wallace, Achille, ??P8 19:06:07 <Zakim> On IRC I see schneid, alanr, msmith, christine, sandro, baojie, zwu21, RRSAgent, Achille, ivan, MarkusK_, Zakim, bmotik, IanH, bijan, ewallace, trackbot 19:06:15 <bijan> zakim, unmute me 19:06:15 <Zakim> bijan should no longer be muted 19:06:16 <zwu21> alanr: do we agree to a simultaneous publication of all docs? 19:06:20 <alanr> ack bijan 19:06:21 <christine> zakim, ??P8 is christine 19:06:22 <Zakim> +christine; got it 19:06:23 <alanr> q+ mike 19:06:50 <alanr> q? 19:07:02 <sandro> Bijan: In a Second-Last-Call, you ask for comments on specifically what has changed. 19:07:08 <zwu21> bijan: my only concern is we need to be careful about second LC is a new round of major comments... 19:07:31 <zwu21> ... otherwise, I am ok with it 19:07:37 <bijan> zakim, mute me 19:07:37 <Zakim> bijan should now be muted 19:08:03 <alanr> q? 19:08:06 <alanr> ack ianh 19:08:15 <bijan> Er...I won't agree to simultaneous unless this is resolved...so I don't see how we can get agreement of simultaneous without the resolution 19:08:16 <zwu21> IanH: I have the same worry as bijan, a second LC gives people chance more comments that may slow down WG progress 19:08:18 <alanr> q+ alanr 19:09:11 <alanr> ack mike 19:09:46 <zwu21> Ivan: by CR, all should be in sync 19:10:01 <zwu21> Mike: want to clarify the consequences 19:10:02 <alanr> q+ sandro 19:10:05 <alanr> ack alanr 19:10:31 <zwu21> alanr: not so worried by TopQuadrant, don't think WG has spent too much time on reponses 19:10:50 <zwu21> ... we can do the same thing if they come back 19:11:01 <bijan> q+ 19:12:10 <zwu21> sandro: your concern about Profiles is editorial, so it can be post LC 19:12:15 <alanr> q? 19:12:18 <alanr> ack sandro 19:12:22 <ivan> ack sandro 19:12:45 <zwu21> ... main point of second LC is the whole package 19:13:05 <zwu21> ... all the rec track spec will be LC, 19:13:25 <bijan> zakim, unmute me 19:13:25 <Zakim> bijan should no longer be muted 19:13:26 <alanr> ack bijan 19:13:26 <zwu21> ... ok with this strategy 19:13:43 <zwu21> bijan: I did not understand Sandro's story 19:13:48 <sandro> sandro: story of LC2 would be "now you get to see the whole package together" 19:14:18 <alanr> q+ sandro 19:14:50 <ivan> zakim, who is here? 19:14:50 <Zakim> On the phone I see MIT346, bijan, Evan_Wallace, Achille, christine 19:14:51 <Zakim> On IRC I see schneid, alanr, msmith, christine, sandro, baojie, zwu21, RRSAgent, Achille, ivan, MarkusK_, Zakim, bmotik, IanH, bijan, ewallace, trackbot 19:14:54 <alanr> q+ schneid 19:14:57 <alanr> ack sandro 19:15:17 <zwu21> sandro: one of the reason is publilsing document review without other documents is strange 19:15:37 <zwu21> ... the story is not perfect, but good enough 19:15:37 <alanr> q+ alanr to ask if there are editor drafts between lc and cr 19:15:37 <bijan> no 19:15:46 <alanr> q+ boris 19:15:53 <zwu21> ... the roadmap will look really odd without other documents 19:15:54 <ivan> ack schneid 19:16:02 <sandro> sandro: the roadmap is screwey if it's linking to 4-months old documents. 19:16:12 <alanr> q+ sandro 19:16:21 <alanr> q+ mike 19:16:22 <zwu21> schneid: we make editorial, we also make design changes. 19:16:55 <bijan> If there's no change to the design? 19:17:09 <schneid> schneid: I don't know whether, e.g., the changes to to the functional syntax and the effects on other documents will /necessarily/ demand a new LC 19:17:12 <alanr> ack alanr 19:17:12 <Zakim> alanr, you wanted to ask if there are editor drafts between lc and cr 19:17:24 <alanr> ack boris 19:17:31 <zwu21> bmotik: I think changes are significant 19:17:34 <bijan> They change implementations :( 19:17:39 <alanr> q+ ianh 19:18:00 <alanr> q+ schneid 19:18:08 <alanr> ack sandro 19:18:55 <alanr> ack mike 19:18:58 <sandro> sandro: second-last-call is required if the positive-reviews would be invalidated 19:18:59 <ivan> ack mike 19:19:08 <bijan> zakim, mute me 19:19:08 <Zakim> bijan should now be muted 19:19:14 <alanr> q+ alanr to mention some substantive changes, e.g. to property chains 19:19:17 <ivan> q+ 19:19:24 <alanr> ack ianh 19:19:24 <ivan> ack ianh 19:19:31 <zwu21> Mike: if we think we need comments on the changes we make, 2nd LC is in order 19:19:47 <ivan> ack schneid 19:19:47 <alanr> ack schneid 19:20:36 <ivan> ack alanr 19:20:36 <Zakim> alanr, you wanted to mention some substantive changes, e.g. to property chains 19:20:36 <zwu21> schneid: to Boris, for 2nd LC, if there is a comment already made, 19:20:52 <zwu21> ... in 1st LC, then we can do minimal 19:21:03 <alanr> ack ivan 19:21:32 <zwu21> ivan: getting beyond LC does not mean it is over 19:21:54 <zwu21> ... in some way, I prefer to have comments now instead of at PR phase 19:22:11 <bijan> There's some advantage to having comments after CR, since we have implementation valdiation 19:22:18 <zwu21> alanr: publish date 3/31/09... 19:23:09 <alanr> q? 19:23:13 <IanH> q? 19:23:14 <zwu21> schneid: end of march should be enough for RDF semantics 19:23:19 <christine> 3/31/09 for UF docs as well ? 19:23:24 <zwu21> ivan: what about Primer, NF&R 19:23:33 <bijan> Primer is fine for another draft by then 19:23:46 <schneid> schneid: end of march will be clearly enough for RDF-Based Semantics 19:23:59 <jar> jar has joined #owl 19:24:14 <alanr> ack ianh 19:24:21 <IanH> q? 19:24:24 <IanH> q+ 19:24:31 <christine> can you write what said about NF&R 19:24:51 <zwu21> jie: 1 month is enough for quick reference 19:25:14 <zwu21> ... the missing links are primer and syntax, 19:25:53 <ewallace> Who will be working on the Primer? 19:26:00 <christine> +q 19:26:01 <zwu21> Markus: end of March is too tight 19:26:03 <Zakim> -bijan 19:26:09 <alanr> q+ pfps 19:26:13 <alanr> q+ ivan 19:26:22 <ivan> q- 19:26:32 <alanr> ack IanH 19:26:58 <alanr> q+ sandro 19:26:58 <zwu21> IanH: if LC is April, Aug will be CR, Oct will be PR, Nov/Dec will be rec 19:27:24 <zwu21> ... and we already said that we want to finish by Dec 19:27:36 <zwu21> ... I want to whole timeline be examined 19:27:39 <alanr> q+ schneid 19:27:42 <zwu21> ... for feasibility 19:27:43 <ivan> q+ 19:27:51 <alanr> ack christine 19:28:01 <pfps> pfps has joined #owl 19:29:35 <ewallace> Ah, now it is clear. 19:29:37 <zwu21> ivan: the NF&R can move directly from LC to PR 19:29:53 <zwu21> christine: is there lots of work to do? 19:30:20 <zwu21> ... maybe NF&R can finish in 1 month as well? 19:30:23 <alanr> ack pfps 19:30:31 <IanH> q? 19:30:33 <zwu21> alanr: we will review it and see what needs to be done 19:30:35 <alanr> ack sandro 19:30:43 <sandro> Editors Done - March 17; begin WG review 19:30:43 <sandro> LC2 published March 1, comment deadline march 29 19:30:43 <sandro> 4-8 weeks handling LC2 comments 19:30:43 <sandro> CR (LC for User Docs), in May 19:30:43 <sandro> PR for everything (but Notes) in July 19:30:44 <sandro> Rec in September 19:31:04 <christine> +q 19:31:29 <zwu21> s/March 1/April 1/g 19:31:36 <IanH> q? 19:31:42 <ivan> ack schneid 19:31:50 <dlm> dlm has joined #owl 19:31:55 <zwu21> schneid: I can finish in the first half of march 19:32:04 <alanr> ack schneid 19:32:15 <zwu21> ... however, what does 2 weeks buy us? 19:32:43 <alanr> ack ivan 19:33:07 <Zakim> + +1.518.276.aaaa 19:33:25 <alanr> q+ ianh 19:33:26 <IanH> q? 19:33:35 <alanr> ack christine 19:34:25 <alanr> ack ianh 19:34:33 <ivan> ack IanH 19:34:54 <alanr> q? 19:35:05 <zwu21> IanH: I appreciate that RDF semantics has to go through LC, 19:35:13 <ivan> ivan: to the question of Christine, the plan is to publish _all_ documents (ie, including quick ref and features) on the same day 19:35:15 <Zakim> +??P14 19:35:24 <zwu21> ... it seems to me that because schedule is tight, 19:35:26 <bijan> zakim, ??p14 is me 19:35:26 <Zakim> +bijan; got it 19:35:34 <zwu21> ... we may want to avoid 2nd LC 19:35:48 <sandro> 09 March - FPWD Document Overview 19:35:48 <sandro> 30 March - Editors Done, begin WG review 19:35:48 <sandro> 13 April - Publish Round 5 (LC2) 19:35:48 <sandro> 01 June - CR 19:35:48 <sandro> 01 Aug - PR 19:35:49 <sandro> 01 Oct - Rec 19:36:10 <bijan> When would CR end? 19:36:18 <bijan> zakim, mute me 19:36:18 <Zakim> bijan should now be muted 19:36:27 <sandro> CR ends 15 July 19:37:02 <alanr> q? 19:37:05 <zwu21> ivan: how long does implenters need for CR to do implementation 19:37:13 <bijan> zakim, unmute me 19:37:13 <Zakim> bijan should no longer be muted 19:37:15 <zwu21> IanH: Pellet and HermiT are very close 19:37:20 <christine> +q 19:37:28 <zwu21> ... HermiT is more or less complete 19:37:43 <zwu21> bijan: Pellet is tracking OWL 2 19:37:56 <ivan> q+ 19:38:03 <alanr> q+ alanr 19:38:12 <bijan> zakim, mute me 19:38:12 <Zakim> bijan should now be muted 19:38:52 <bijan> FPWD, publish early and often 19:38:53 <alanr> ack christine 19:38:56 <alanr> ack ivan 19:39:04 <zwu21> christine: can we set is to Mar 9? 19:39:12 <pfps> pfps has joined #owl 19:39:25 <alanr> q+ pfps 19:39:37 <christine> scan we set is to Mar 9?/ 15 19:39:57 <ivan> ack alanr 19:39:59 <zwu21> ivan: the LC version can have your current comments 19:40:14 <bijan> I have more answers 19:40:23 <zwu21> ... question of Profiles implementation 19:40:27 <alanr> q+ pfps 19:40:32 <alanr> q+ mike 19:40:36 <alanr> ack pfps 19:40:52 <zwu21> pfps: HermiT is an implementation complete for everything except for syntax checking 19:41:10 <christine> for scribe : christine asked : can we set is to Mar 15 not 9 19:41:17 <zwu21> ... given a RL document, it will do RL reasoning 19:41:54 <alanr> ack mike 19:41:56 <alanr> q+ pfps 19:41:59 <bijan> q+ 19:42:13 <zwu21> Mike: if we have Pellet and Hermit, then we have 2 implementations 19:42:21 <alanr> q+ alanr to ask whether hermit is an "spirit of the law" implementation of RL 19:42:35 <zwu21> ... Pellet RC can support RL and QL 19:42:59 <alanr> ack pfps 19:43:02 <zwu21> ivan: no RL implementation 19:43:13 <zwu21> pfps: what do we need for CR exit status 19:43:18 <alanr> q+ schneid 19:43:22 <zwu21> ... I don't think we need a product 19:43:37 <zwu21> sandro: two interoperable implementations 19:43:38 <alanr> q+ boris 19:43:59 <alanr> q+ ianh 19:44:04 <alanr> ack bijan 19:44:04 <bijan> zakim, unmute me 19:44:06 <sandro> ack bijan 19:44:07 <Zakim> bijan should no longer be muted 19:44:54 <zwu21> bijan: regarding profiles, for QL, there are 3 19:45:04 <sandro> Bijan: QL implementations: C&P, Aberdeen, Rome 19:45:07 <zwu21> ... for EL, IBM has one 19:45:08 <alanr> q+ pfps 19:45:12 <Achille> q+ 19:45:13 <baojie> q+ 19:45:18 <pfps> q- 19:45:37 <zwu21> ... for profile checkers, there will be one from Machnester 19:45:44 <zwu21> ... one from Aberdeen 19:46:11 <ivan> ack alanr 19:46:11 <Zakim> alanr, you wanted to ask whether hermit is an "spirit of the law" implementation of RL 19:46:18 <alanr> ack schneid 19:46:22 <alanr> ack boris 19:46:43 <zwu21> bmotik: regarding profiles, if it is about an implementation that pass the tests, then do we can about implementation details? 19:46:51 <zwu21> s/can/care 19:46:54 <alanr> q+ pfps 19:46:59 <pfps> q- 19:47:13 <alanr> ack ianh 19:47:18 <zwu21> IanH: I think we already have enough implementations, 19:47:21 <alanr> q+ schneid 19:47:37 <alanr> ack Achille 19:47:50 <zwu21> Achille: want to clarify IBM's implementation of EL++, 19:47:59 <zwu21> ... is a simplfied version 19:47:59 <alanr> ack baojie 19:48:30 <christine> +q 19:48:32 <alanr> ack schneid 19:48:42 <alanr> q+ 19:48:47 <zwu21> schneid: CR's purpose is to find bugs and implementation difficulty, 19:48:54 <zwu21> ... now, we already have enough 19:49:14 <alanr> q+ ianh 19:49:20 <alanr> ack christine 19:49:21 <ivan> ack christine 19:49:46 <zwu21> alanr: it is not necessary to delay because we want to keep an schedule 19:50:05 <zwu21> ... you know what, let us communicate in emails 19:50:40 <bijan> +1 to ivan, publishing wds *should be cheap* 19:50:57 <bijan> What's the question? 19:51:01 <ewallace> What is the question? 19:51:32 <alanr> q? 19:52:03 <zwu21> christine: want to understand why it is hard to set the date 15th 19:52:05 <bijan> There's a schedule and there's no real benefit. FPWD is a low bar 19:52:15 <zwu21> alanr: happy to discuss offlien 19:52:22 <zwu21> s/offlien/offline/g 19:53:04 <zwu21> ivan: from CR to PR, we come up with a report on implementations 19:53:40 <zwu21> alanr: do we expect comments on PR? 19:54:17 <zwu21> ivan: it is possible, that is why I want comments now 19:54:33 <zwu21> ... not on PR documents 19:55:04 <zwu21> IanH: if schedule slips, then it is going to be tight for dec 2009 19:55:29 <alanr> ? 19:55:31 <alanr> q? 19:55:35 <alanr> ack alanr 19:55:38 <alanr> ack inah 19:55:43 <alanr> ack ianh 19:56:01 <pfps> I'm not happy with the schedule, but it is about as good as it could be 19:56:02 <zwu21> sandro: we chould consider skip CR 19:56:16 <pfps> However, we should use the schedule as a cloture mechanism 19:56:49 <zwu21> ivan: let us not skip CR 19:57:37 <zwu21> ... what we called user facing documents do not go through CR 19:57:38 <sandro> 09 March - Publich Round 5: FPWD Document Overview 19:57:38 <sandro> 30 March - Editors Done, begin WG review 19:57:38 <sandro> 15 April - Publish Round 6: All documents, specs in Last Call (LC1 or LC2) 19:57:38 <sandro> 01 June - Publish Round 7: All docs; rec-track specs to CR 19:57:38 <sandro> 15 July - CR comments due 19:57:39 <sandro> 01 Aug - Publish Round 8: All docs; rec-track documents to PR 19:57:42 <sandro> 01 Oct - Publish Round 9: All documents to final state (Rec / Note) 19:57:43 <zwu21> ... that gives up more time 19:57:57 <zwu21> ... Manchester syntax does not go through CR because it is not rec track 19:58:10 <zwu21> ... if it is final, we can publish it as a note anytime 19:58:19 <bijan> We shouldn't solicit comments on a note 19:58:29 <sandro> http://www.w3.org/2007/OWL/wiki/Timeline 19:58:48 <bijan> The only reason to go not go final on MS now is to track any changes we make in the rest of the langauge 20:00:13 <zwu21> ivan: at PR, we may get formal objections 20:00:28 <zwu21> ... which will be a very tough thing 20:00:48 <zwu21> Mike: it is indepenent of our timeline though 20:00:52 <IanH> PROPOSED: the WG will use its best endeavours to complete its work according to the schedule proposed by Sandro above. 20:01:02 <christine> +q 20:01:17 <IanH> q? 20:01:20 <alanr> ack christine 20:02:02 <zwu21> christine: don't see the impact of either 9th of 15th 20:02:07 <zwu21> alanr: we will address that 20:02:25 <pfps> +1 ALU 20:02:31 <alanr> +1 SC 20:02:32 <schneid> +1 20:02:32 <ivan> +1 20:02:32 <IanH> +1 20:02:33 <bmotik> +1 20:02:33 <MarkusK_> +1 FZI 20:02:33 <zwu21> +1 20:02:34 <bijan> +1 20:02:36 <msmith> +1 20:02:37 <sandro> +1 20:02:38 <Achille> +1 20:02:40 <baojie> +1 20:02:59 <christine> +1 (except 09 march) 20:03:00 <zwu21> alanr: we need to note that which documents will go to LC2 20:03:06 <ewallace> +1 20:03:26 <IanH> RESOLVED: the WG will use its best endeavours to complete its work according to the schedule proposed by Sandro above. 20:03:42 <bijan> Is all that's left editorial? 20:03:50 <bijan> I have a course to prepare for and to go to sleep :( 20:04:04 <pfps> no, after break is imports and griddle 20:04:28 <Zakim> -christine 20:08:22 <Zakim> - +1.518.276.aaaa 20:08:28 <Zakim> -Evan_Wallace 20:17:51 <bijan> zakim, mute me 20:17:51 <Zakim> bijan should now be muted 20:19:30 <jar> jar has joined #owl 20:20:26 <zwu21> IanH: going to imports 20:20:58 <pfps> scribenick: ivan 20:21:08 <pfps> it should be one 20:21:25 <ivan> Topic: imports 20:21:37 <ivan> IanH: 2 comments, both with drafts 20:21:41 <ivan> ... #53 20:22:22 <IanH> http://lists.w3.org/Archives/Public/public-owl-comments/2009Feb/0000.html 20:22:36 <pfps> draft response is http://www.w3.org/2007/OWL/wiki/LC_Responses/JR7 20:22:55 <ivan> IanH: proposed draft: 20:23:35 <sandro> sorry, my IP address changed on me. webcam restarted.... 20:23:39 <IanH> q? 20:23:53 <jar> q+ jar 20:23:57 <ivan> schneid: there are confusions due to sloppiness of myself 20:24:12 <ivan> ... in the old owl 1 full ther ehas been a definition of import closure 20:24:36 <IanH> q? 20:24:48 <ivan> ... i kept that in in the 2nd draft 20:24:58 <baojie> q+ 20:24:59 <ivan> ... there was also a note that was very clever; 20:25:09 <ivan> ... ie, i did not plan to have this in the final version of the document 20:25:26 <ivan> ... in the owl 1 the definition was only used in two theorems 20:25:47 <ivan> ... on of them was the old correspondence theorem, and there is a new one for owl 2 that does not use this any more 20:25:52 <IanH> q? 20:25:53 <ivan> s/on/one/ 20:26:01 <ivan> .... my current draft does not have it any more 20:26:22 <ivan> IanH: ie,, the current version of the owl full semantics does not have this feature in 20:26:29 <ivan> schneid: indeed 20:26:46 <ivan> ... importing has nothing to do with logic, treating it in a semantics is not correct 20:27:12 <Zakim> +Evan_Wallace 20:27:16 <IanH> q? 20:27:25 <IanH> ack jar 20:27:34 <schneid> http://www.w3.org/2007/OWL/wiki/RDF-Based_Semantics#Content_of_Ontologies_.28Informative.29 20:27:34 <ivan> jar: since i submitted that I was thinking about it. 20:27:40 <ivan> ... this is a borderline editorial 20:27:49 <ivan> ... i am not sure what the goal for today 20:27:58 <ivan> ... i guess it is the lc comments 20:28:18 <ivan> ... i did sent another public comment today on how to present this whole comment idea 20:28:37 <ivan> ... i am happy to contribute and work with whoever works on this 20:28:57 <ivan> IanH: you should send a mail to the wg list targeted at michael, and then discuss this 20:29:01 <ivan> ... is that o.k? 20:29:07 <ivan> jar: yes, that sounds fine 20:29:10 <IanH> q? 20:29:19 <jar> my email (today): http://lists.w3.org/Archives/Public/public-owl-comments/2009Feb/0033.html 20:29:35 <ivan> ... I feel there is a lot of room for improvement 20:29:52 <ivan> IanH: but if the response to you was along the line that this document is chaning 20:29:55 <IanH> q? 20:29:55 <ivan> ... is that ok 20:29:59 <ivan> jar: yes 20:30:38 <ivan> baojie: about the semantics of incompatibility with in owl 1 we do not have that, so we have a backward incompatibility problem 20:30:50 <jar> any clarification is fine I think. just wanted to make sure someone had thought about it, and that the next reader was clear on the intent (full different from / same as dl in this way) 20:30:58 <IanH> q? 20:31:01 <ivan> IanH: i am reluctant to reopen this 20:31:05 <IanH> ack baojie 20:31:17 <ivan> boris: michael you defer to the syntax document? 20:31:44 <ivan> schneid: i have either an own part that treats this stuff or not, i decided to point to the syntax document 20:31:57 <ivan> bmotik: I agree 20:32:12 <ivan> IanH: we are done on this one, aren't we? 20:32:22 <ivan> ... do we have a response draft? 20:32:48 <ivan> pfps: (reads up the response) 20:33:09 <ivan> IanH: the response is that this is not last call, the document has changed, the import is not a semantics operation 20:33:24 <ivan> ... further efforts will be made to improve the presentation 20:34:21 <bijan> q+ 20:34:22 <ivan> Topic: import by location 20:34:27 <bijan> zakim, unmute me 20:34:27 <Zakim> bijan should no longer be muted 20:34:29 <IanH> q? 20:34:32 <IanH> ack bijan 20:35:13 <ivan> bijan: we had an extensive discussion with Tim ???, Peter has a very long and involved response and we trim that 20:35:21 <bijan> zakim, mute me 20:35:21 <Zakim> bijan should now be muted 20:35:23 <IanH> q? 20:35:27 <bijan> I happy to 20:35:29 <ivan> pfps: I volunteer bijan to write it:-) 20:35:58 <bijan> I head to the page 20:36:09 <ivan> Topic: 2 comments on axiom annotation 20:36:26 <ivan> IanH: they are from bijan, asking for axiom hiding and for naming 20:36:29 <bijan> zakim, unmute me 20:36:29 <Zakim> bijan should no longer be muted 20:36:32 <ivan> ... i though we agreed 20:36:38 <ivan> ... and the commenter agreed, too 20:36:57 <ivan> bijan: i think we decided to reject that and i had the action to answer to myself 20:37:04 <ivan> pfps: i can do it 20:37:10 <ivan> ... i know how to abuse you nicely 20:37:52 <ivan> Topic: number 30, frank's objection 20:38:18 <IanH> Ivan: discussed this to death 20:38:26 <bijan> ship it! 20:38:28 <IanH> Ivan: version on the web agrees with discussion 20:38:40 <msmith> http://www.w3.org/2007/OWL/wiki/LC_Responses/FH4 20:39:19 <bijan> I just reread it and it's great! 20:39:31 <IanH> PROPOSED: send drafted response to comment 30 20:39:34 <ewallace> +1 20:39:36 <IanH> +1 20:39:36 <ivan> +1 20:39:38 <zwu21> +1 20:39:40 <pfps> +1 ALU 20:39:52 <schneid> +1 20:39:54 <baojie> 0 20:39:59 <msmith> +1 20:40:01 <MarkusK_> +1 20:40:04 <bmotik> +1 20:40:06 <sandro> +1 20:40:09 <alanr> +1 20:40:11 <IanH> RESOLVED: send drafted response to comment 30 20:40:14 <bijan> +1 20:40:29 <pfps> http://lists.w3.org/Archives/Public/public-owl-comments/2009Feb/0005.html 20:40:36 <ivan> Topic: number 58, strong typing 20:40:47 <ivan> pfps: the answer is yes, was part of the discussion yesterday 20:41:03 <ivan> IanH: it brings the fs and the structure aligned 20:41:20 <ivan> bmotik: I will do it, 'thank you, we will do it' 20:41:42 <bijan> We discussed this yesterday 20:41:57 <bijan> I have and action to send the schema (nearly done :() 20:42:20 <ivan> Topic: number 47, disallow multiple key values 20:42:31 <ivan> IanH: no multiple key values 20:42:38 <pfps> http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0062.html 20:43:14 <bijan> q+ 20:43:20 <ivan> pfps: I suggest to say no to this 20:43:22 <IanH> q? 20:43:39 <ivan> schneid: talking to database people they say this is plainly wrong 20:43:42 <bijan> zakim, unmute me 20:43:42 <Zakim> bijan was not muted, bijan 20:43:44 <IanH> ack bijan 20:44:02 <ivan> bijan: i agree with schneid and peter 20:44:10 <ivan> ... you can get that if you wanted 20:44:15 <bijan> zakim, mute me 20:44:15 <Zakim> bijan should now be muted 20:44:30 <ivan> pfps: i will take it 20:44:51 <IanH> http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0048.html 20:44:53 <ivan> Topic: automatic testing in the owl link interface (45) 20:45:15 <bijan> "Thanks for the comment." 20:45:26 <ivan> IanH: i think our response is that it is out of scope 20:45:47 <bijan> q+ 20:45:58 <ivan> pfps: uli is on the hook for that one 20:46:32 <ivan> pfps: ship it! 20:46:37 <bijan> q- 20:46:42 <IanH> q? 20:47:04 <IanH> PROPOSED: send response as drafted to comment 45 20:47:06 <pfps> +1 ALU 20:47:08 <ivan> 0 20:47:12 <IanH> +1 20:47:13 <ewallace> +1 20:47:17 <Achille> Achille has joined #owl 20:47:55 <bijan> +1 20:48:05 <ivan> msmith: one problem, the text says that we will publish document and test cases, but that is not exactly true 20:48:14 <ivan> ... we will produce a test collection 20:48:52 <zwu21> +1 20:49:03 <msmith> +1 20:49:05 <MarkusK_> 0 20:49:13 <schneid> +1 20:49:16 <IanH> RESOLVED: send response as drafted to comment 45 20:49:21 <MarkusK_> I meant +1 20:49:28 <Achille> +1 20:49:45 <ivan> Topic: number 23, extending annotation 20:49:57 <IanH> http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0028.html 20:50:02 <ivan> pfps: after a long discussion with jeremy roger he and i approved a response 20:50:10 <IanH> http://www.w3.org/2007/OWL/wiki/LC_Responses/JR1 20:50:26 <ivan> pfps: "we would love to do, but nobody knows how@ 20:50:31 <ivan> s/@/"/ 20:50:34 <bijan> I've drafted a response to TR1: http://www.w3.org/2007/OWL/wiki/LC_Responses/TR1 20:50:45 <IanH> Q? 20:50:47 <bijan> sorry, jsut reporting 20:50:48 <bijan> not relevant 20:51:13 <IanH> PROPOSED: send draft response http://www.w3.org/2007/OWL/wiki/LC_Responses/TR1 20:51:26 <bijan> Not tr1 20:51:30 <bmotik> +1 20:51:31 <pfps> +1 20:51:32 <ivan> 0 20:51:33 <msmith> +1 20:51:34 <IanH> +1 20:51:35 <MarkusK_> +1 20:51:36 <bijan> no nono 20:51:36 <zwu21> +1 20:51:39 <schneid> +1 20:51:40 <bijan> zakim, unmute me 20:51:40 <Zakim> bijan should no longer be muted 20:51:48 <baojie> +1 20:52:22 <IanH> PROPOSED: send draft response http://www.w3.org/2007/OWL/wiki/LC_Responses/JR1 20:52:24 <msmith> +1 20:52:27 <bijan> +1 20:52:28 <pfps> +1 alu 20:52:30 <bijan> zakim, mute me 20:52:30 <Zakim> bijan should now be muted 20:52:32 <alanr> +1 20:52:32 <zwu21> +1 20:52:50 <IanH> RESOLVED: send draft response http://www.w3.org/2007/OWL/wiki/LC_Responses/JR1 20:52:52 <ewallace> 0 20:53:05 <bijan> zakim, mute me 20:53:05 <Zakim> bijan was already muted, bijan 20:53:19 <bijan> zakim, unmute me 20:53:19 <Zakim> bijan should no longer be muted 20:53:25 <ivan> Topic: number 7, import via 20:53:41 <IanH> PROPOSED: send draft response http://www.w3.org/2007/OWL/wiki/LC_Responses/TR1 20:53:42 <ivan> bijan: just put up a response 20:54:07 <pfps> +1 20:54:10 <bijan> +1 20:54:14 <schneid> +1 20:54:18 <zwu21> +1 20:54:25 <ivan> +1 20:54:27 <ewallace> +1 20:54:31 <baojie> +1 20:54:31 <Achille> +1 20:54:32 <MarkusK_> +1 20:54:36 <bmotik> +1 20:54:40 <msmith> +1 20:54:45 <alanr> +1 20:54:45 <IanH> RESOLVED: send draft response http://www.w3.org/2007/OWL/wiki/LC_Responses/TR1 20:57:24 <sandro> http://www.w3.org/2007/OWL/wiki/LC_Responses/JC1a says GRDDL 20:57:24 <sandro> Our understanding of the WG charter is that a GRDDL transform, in XSLT1, will be provided. We will raise this issue again at PR review if necessary. Our preferred fix to the lack of a GRDDL transform, is to drop the OWL/XML serialization. 20:57:45 <ewallace> Have to go for an hour. I will be back at 5, if you are still meeting. 20:57:48 <Zakim> -Evan_Wallace 20:58:05 <ivan> Topic: number 17, 34 plus a bunch together, GRDDL 20:58:49 <ivan> IanH: 17 says that there is an open issue, will that be resolved? 20:58:51 <bijan> I am 20:58:52 <IanH> q? 20:58:54 <Zakim> -Achille 20:58:58 <bijan> zakim, unmute me 20:58:58 <Zakim> bijan was not muted, bijan 20:59:15 <ivan> bijan: i have seen no change, nobody has talked to me, 20:59:47 <ivan> sandro: last i remember (last f2f) we might move forward with a plan with a grddl that would get to a transform 21:00:00 <ivan> bijan: we got a push back from jonathan 21:00:52 <sandro> ivan: I am unsure whether it is doable in XSLT. 21:00:54 <bijan> It doesn't seem that TQ would be happy with it either 21:00:58 <AchilleF> AchilleF has joined #owl 21:01:08 <bijan> q+ 21:01:15 <ivan> jar: i think it is clear you can do it in xslt, so it is a question of service 21:01:16 <dlm> dlm has joined #owl 21:01:25 <ivan> ... doing it relying on service is a bit fragile 21:01:27 <sandro> jar: I think you CAN do it in XSLT. It's a question of judgement. I think relying on a service is really quite fragile, as opposed to relying on a program. 21:01:34 <ivan> ... the argument is that it is the same sort of thing 21:01:39 <Zakim> +[IBM] 21:01:45 <bijan> Not catchign everything 21:01:45 <IanH> q? 21:01:49 <bijan> Could he move closer to the phone 21:01:51 <ivan> ... and it relies on a service 21:01:51 <AchilleF> zakim, ibm is me 21:01:51 <Zakim> +AchilleF; got it 21:01:52 <ivan> q+ 21:02:11 <IanH> q? 21:02:25 <ivan> ... i think it can be done in many different ways, one way is a service, i am advocating for a proof of something more robust 21:02:34 <IanH> q+ sandro 21:02:39 <ivan> ... how hard is it to replicate this transform, can I copy the software, etc 21:02:41 <bijan> zakim, unmute me 21:02:41 <Zakim> bijan was not muted, bijan 21:02:43 <IanH> q+ peter 21:02:56 <IanH> ack bijan 21:03:12 <sandro> bijan: associating a URI with a mapping is sufficient, I believe. 21:03:26 <ivan> bijan: my original point was that a URI pointing at a generic thing is enough 21:03:38 <IanH> q? 21:03:56 <alanr> q+ 21:04:01 <sandro> bijan: then we were asked for an XSLT, and that proves you're really asking for a program, not a spec. 21:04:04 <pfps> bijan: we are heading down the slippery slope to implementation 21:04:05 <ivan> scribe gave up scribing bijan 21:04:12 <IanH> q? 21:04:18 <pfps> bijan: let's do something minimal 21:04:19 <bijan> zakim, mute me 21:04:19 <Zakim> bijan should now be muted 21:04:26 <IanH> ack ivan 21:04:44 <alanr> q- 21:04:44 <alanr> q- alanr 21:05:18 <IanH> q? 21:05:22 <sandro> ivan: I don't want to reopen this. We agree to disagree. The problem (cf JAR) -- we have no one producing that XSLT that converts OWL/XML to RDF/XML. If so, then we could talk to Bijan about it. But we don't have it. 21:05:24 <dlm> dlm has joined #owl 21:05:53 <sandro> ivan: Will you make the implementation JAR? 21:06:04 <sandro> jar: No. 21:06:11 <alanr> q+ 21:06:19 <IanH> ack sandro 21:06:20 <bijan> What? Quality? Huh? 21:06:57 <jar> to reflect back what bijan said: the grddl uri 'identifies' not a script or program, but the transformation. any implementation of it that works is fine; and implementation is outside the scope of the spec. 21:07:00 <ivan> sandro: bijan you talked about another possibility if the xslt were produced mechanically and if necessary i might volounteer to do taht 21:07:10 <bijan> zakim, unmute me 21:07:10 <Zakim> bijan should no longer be muted 21:07:10 <IanH> q? 21:07:11 <alanr> q? 21:07:14 <bijan> q+ 21:07:15 <ivan> ... i hear bijan say that he is against that 21:07:52 <ivan> pfps: in sympathy with bijan here, bijan's solution is to reuse another tool that will go through our cr tool, 21:07:56 <IanH> q? 21:08:01 <bijan> Example: http://owl.cs.manchester.ac.uk/converter/ 21:08:02 <ivan> ... and plan it to be make it available 21:08:04 <IanH> ack peter 21:08:13 <ivan> ... it is code reuse, which is good 21:08:22 <ivan> ... the only thing it does not have is normativity 21:08:32 <IanH> q+ sandro 21:08:35 <ivan> ... if you wan normativity to point to our document 21:08:49 <ivan> ... then there is no code, no viruses, no nothing... 21:08:55 <IanH> ack alanr 21:09:19 <IanH> q+ 21:09:21 <ivan> alanr: i have already scraped the document once and that can be an input to sandro's script 21:09:50 <ivan> ... fair to say that publishing a spec without is a minority view 21:09:59 <bijan> zakim, unmute me 21:09:59 <Zakim> bijan was not muted, bijan 21:10:01 <IanH> ack bijan 21:10:05 <sandro> alan: It's a minority view that spec or on-line service is "grddl" per se. 21:10:13 <ivan> ... we may have a formal objection that we have to consider, do the damn thing and let it done 21:10:22 <sandro> alan: It's a minority view that spec is "grddl" per se. 21:10:47 <sandro> bijan: My main objection is to on-line downloadability. 21:11:13 <pfps> q? 21:11:16 <ivan> bijan: i disagree with what alan says, my main objection is a downloadable script, having a page with a set of transformation is fine, manchester might put an objection if we do thi 21:11:17 <pfps> q+ 21:11:35 <IanH> ack sandro 21:11:49 <ivan> bijan: i have in principle objections the way grddl work 21:11:56 <ivan> ... maybe the mechanical would work 21:12:03 <ivan> ... maybe we should just table this 21:12:05 <sandro> bijan: the mechanical-generation of the transform doesn't help. 21:12:09 <ivan> ... i cannot promise i will agree 21:12:09 <IanH> q? 21:12:10 <bijan> zakim, mute me 21:12:11 <Zakim> bijan should now be muted 21:12:25 <bijan> zakim, unmute me 21:12:25 <Zakim> bijan should no longer be muted 21:12:25 <ivan> sandro: a few weeks i do not want to spend unless i need to 21:12:27 <bijan> q+ 21:12:49 <ivan> ... i do not know whether we can judge the strength of the objections 21:13:07 <IanH> q+ msmith 21:13:17 <ivan> bijan: proposal, if sandro produces such a thing and it pasts all the tests, then i have a strong bias to accept it 21:13:21 <sandro> bijan: if you produce such a thing, and it passes all the tests, I will have a strong bias in favor of supporting it, although I can't promise I'll accept it. 21:13:26 <IanH> q? 21:13:35 <sandro> thanks, Bijan. 21:13:45 <bijan> zakim, mute me 21:13:45 <Zakim> bijan should now be muted 21:14:22 <bijan> Or defaultly use 21:14:24 <ivan> IanH: bijan seems to say that he has an in principle objection to specify a piece of software 21:14:41 <ivan> ... to be honest i fully sympathize with that objection 21:14:48 <ivan> sandro: no one is saying 21:15:18 <IanH> q? 21:15:35 <IanH> ack IanH 21:15:47 <bijan> It's de facto have to use otherwise this wouldn't matter 21:15:51 <bijan> It's defaultly used 21:15:58 <alanr> there is no "have to" anywhere 21:16:20 <sandro> sandro: all the spec says is "use this namespace". then it's up to the namespace owner (W3C, guided by the WG) to make sure the right (GRDDL) thing happens. 21:16:33 <bijan> as a *Second* specification! 21:17:11 <IanH> q? 21:17:18 <schneid> and if we simply put silently, without spec'ing it, a GRDDL transform at the OWL URL? :-) 21:17:18 <bijan> q- 21:17:53 <pfps> we annoint the transform by using the namespace 21:18:14 <bijan> The editor, DanC, agreed with my interpretation 21:18:16 <sandro> ivan: Whether we like it or not, there is GRDDL. It's done. That's not for us to discuss. Bijan and I disagree about what that spec means, but.... 21:18:20 <sandro> q? 21:18:55 <sandro> ivan: Bottom-Line: if this is not set up the way the community is set up, there will be formal objections to OWL/XML. 21:18:56 <bijan> THat's why I caved 21:19:02 <IanH> ack pfps 21:19:08 <sandro> ivan: Bottom-Line: if this is not set up the way the community expects it, there will be formal objections to OWL/XML. 21:19:19 <bijan> However, will the director not override the objection? 21:19:21 <bijan> q+ 21:19:43 <IanH> q+ 21:20:43 <bijan> Uhm, I have the editor telling me that my interpretation is correct 21:21:11 <sandro> peter: If you like to XSLT in GRDDL,you're annointing that XSTL as *the* *definition* 21:21:35 <sandro> sandro: no, that's ridiculous. The real spec is still obvious the Recommendation, which the XSLT implements. 21:21:43 <schneid> ian: if we create such an implementation, but explicitly say that it is not our spec, what happens then? 21:21:47 <ivan> q+ 21:21:55 <IanH> ack msmith 21:22:17 <bijan> http://www.w3.org/2004/01/rdxh/spec#txforms 21:22:30 <bijan> As noted above, each GRDDL transformation specifies a transformation property, a function from XPath document nodes to RDF graphs. This function need not be total; it may have a domain smaller than all XML document nodes. For example, use of xsl:message with terminate="yes" may be used to signal that the input is outside the domain of the transformation. 21:22:30 <bijan> Developers of transformations should make available representations in widely-supported formats. XSLT version 1[XSLT1] is the format most widely supported by GRDDL-aware agents as of this writing, though though XSLT2[XSLT2] deployment is increasing. 21:22:35 <schneid> msmith: asks, people, you want a single XSD that would be referenced? 21:22:47 <bijan> Who uses GRDDL? 21:22:53 <schneid> alanr: people who use grddl, should have their expectations met 21:22:57 <bijan> I mean, what's the population? 21:22:59 <sandro> alan: What I think is needed is the kind of thing GRDDL users want/expect. 21:23:02 <IanH> q? 21:23:22 <bijan> zakim, unmute me 21:23:22 <Zakim> bijan should no longer be muted 21:23:25 <IanH> ack bijan 21:23:55 <IanH> q+ pfps 21:24:09 <schneid> bijan: grddl chair made assertion that conflicts with what sandro sais 21:24:13 <pfps> from GRDDL abstract: Abstract 21:24:15 <pfps> GRDDL is a mechanism for Gleaning Resource Descriptions from Dialects of Languages. This GRDDL specification introduces markup based on existing standards for declaring that an XML document includes data compatible with the Resource Description Framework (RDF) and for linking to algorithms (typically represented in XSLT), for extracting this data from the document. 21:24:30 <schneid> bijan: our spec is the document 21:24:45 <schneid> bijan: understands positions of both parties 21:24:51 <IanH> q? 21:25:31 <schneid> bijan: asks, whether ivan suggests to not make the grddl transform a rec? 21:25:36 <sandro> ivan: Of course I can't know what will happen if there is a formal objection. 21:25:50 <IanH> q? 21:25:58 <schneid> ivan: if there is a formal objection, then this will probably kill owl/xml as a rec 21:26:40 <sandro> q? 21:26:44 <bijan> zakim, mute me 21:26:44 <Zakim> bijan should now be muted 21:26:46 <ivan> q- 21:26:53 <sandro> ack IanH 21:27:08 <bijan> users will never see those disclaimers 21:27:18 <sandro> IanH: maybe do the XSLT, and hedge around it with various warnings. 21:27:20 <IanH> q? 21:27:24 <schneid> ianh: no one has commented on my suggestion: "this grddl transform is not a mandatory spec" 21:27:29 <IanH> ack pfps 21:28:39 <bijan> People don't pick it up...it's silent 21:28:39 <schneid> alanr: don't understand what the problem is with having bugs in the transform, then let's fix it; the normative thing is the document 21:28:40 <bijan> q+ 21:29:17 <IanH> q? 21:29:19 <bijan> zakim, unmute me 21:29:19 <Zakim> bijan should no longer be muted 21:29:19 <schneid> alanr: why not document this that we will fix all bugs 21:29:21 <pfps> q+ 21:29:22 <IanH> ack bijan 21:29:38 <IanH> ack pfps 21:29:54 <schneid> bijan: I am not going to accept this [FIXME!] 21:30:11 <IanH> q? 21:30:12 <bijan> Yes 21:30:26 <bijan> q+ to point out a service based example 21:30:48 <bijan> http://inamidst.com/sw/hturtle/ 21:31:15 <IanH> q? 21:31:27 <schneid> ivan: in f2f4 we discussed that there should /exist/ some xslt transform as a service (somehow) 21:31:56 <bijan> zakim, unmute me 21:31:57 <Zakim> bijan was not muted, bijan 21:32:03 <IanH> q? 21:32:18 <IanH> ack bijan 21:32:18 <Zakim> bijan, you wanted to point out a service based example 21:32:22 <IanH> q? 21:32:27 <bijan> We did dicusss it 21:32:37 <bijan> zakim, unmute me 21:32:37 <Zakim> bijan was not muted, bijan 21:32:40 <IanH> ack bijan 21:32:45 <schneid> bijan: there is a conversion service (have put url into irc) 21:33:05 <schneid> ivan: we can have such a service at w3c, too 21:33:24 <IanH> q? 21:33:28 <schneid> alanr: what is the issue with this? 21:33:45 <IanH> q? 21:33:46 <schneid> alanr: how would this solve the problem? 21:33:59 <IanH> q? 21:33:59 <schneid> ivan: does not resolve the principle problem 21:34:11 <bijan> q+ 21:34:21 <bijan> zakim, unmute me 21:34:21 <Zakim> bijan was not muted, bijan 21:34:29 <IanH> q? 21:34:46 <IanH> q? 21:34:49 <IanH> ack bijan 21:35:37 <IanH> q? 21:35:53 <bijan> bijan: The service based one makes me a little happier (not ideal) because it makes it very clear that there is no specification going on with this implementation 21:36:10 <bijan> Like the RDF Mapping 21:36:35 <IanH> q? 21:36:39 <sandro> jar: for practical reasons, the grddl transform must resolve to code that will run. 21:37:08 <schneid> jonathan: issue for me is how to have the transform behind the firewall 21:37:17 <bijan> Why is firewall replication a criterion? 21:37:18 <IanH> q? 21:37:18 <sandro> jar: then the question is how hard will it be for me to implement the transform behind my firewall. (1) punch a hole, (2) copy the service, (3) re-implenet 21:37:28 <sandro> q+ 21:37:29 <bijan> q- 21:37:30 <sandro> q- 21:38:31 <bijan> It is open source 21:38:43 <IanH> q? 21:39:13 <sandro> peter: grddl should selectiveally apply any/all trnasforms. 21:39:15 <IanH> q? 21:39:18 <bijan> Peter is wrong 21:39:30 <bijan> q+ 21:39:54 <IanH> q? 21:40:01 <bijan> zakim, unmute me 21:40:01 <Zakim> bijan was not muted, bijan 21:40:09 <IanH> ack bijan 21:40:20 <pfps> http://www.w3.org/TR/grddl/#sec_agt 21:40:30 <schneid> bijan: belives peter's reading of the grdll spec is invalid 21:40:52 <IanH> q? 21:40:54 <schneid> bijan: one can have several transforms for the same 21:41:02 <bijan> I agree! 21:41:04 <IanH> q? 21:41:15 <IanH> q? 21:41:40 <bijan> I proposed this way back and was rejected :( 21:41:42 <schneid> pfps: if we cannot understand the grddl spec, then it is not perfect 21:41:48 <IanH> q? 21:41:51 <msmith> peter is referencing the second normative (green) block at http://www.w3.org/TR/grddl/#sec_agt 21:41:57 <sandro> alan: nice solution would be to have multiple transforms. 21:42:01 <bijan> The GRDDL chair suggested that this is nice idea 21:42:11 <jar> what I need a story for: What happens 10 years from now after the service stops running (you look at the spec and reimplement it?); what do I do inside the firewall (same? or find the java code?) 21:42:58 <IanH> q? 21:43:14 <schneid> ivan: service oriented solution works for me, and if it also works for jar and tq, then its ok? 21:43:17 <bijan> jar, what would happen if the w3c servers died and no longer served the XSLT? 21:43:41 <jar> go to the time machine. 21:43:52 <schneid> alanr: ivan, you would not object 21:43:54 <schneid> ivan: no 21:43:54 <bijan> So, same deal 21:44:01 <sandro> PROPOSED: if we can have multiple GRDDL transforms, then we'll do (1) the spec, (2) on-line transform service, (3) download single xslt; if only one transform, then we'll do the on-line transform service only. 21:44:49 <sandro> DanC: it's OK to have multiple XSLTs; the GRDDL test suite has an example, yes 21:44:49 <sandro> DanC: if you're willing to claim the spec is a representation of an algorithm, then yes, you can link the spec as a GRDDL transformation. 21:44:52 <pfps> +1 ALU 21:45:02 <ivan> 1 21:45:06 <IanH> ? 21:45:35 <IanH> q? 21:45:35 <sandro> (I'm quoting what Dan just answered me.) 21:46:41 <schneid> sandro: what will have happen, if I don't manage it? 21:46:45 <ivan> 0 21:46:47 <zwu21> 0 21:46:51 <sandro> PROPOSED: if we can have multiple GRDDL transforms, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), (3) download single xslt (if a good one is produced); if only one transform, then we'll do the on-line transform service only. 21:47:05 <sandro> PROPOSED: if we can have multiple GRDDL transforms, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); if only one transform, then we'll do the on-line transform service only. 21:47:26 <bijan> WHy wouldn't we do the spec? 21:48:00 <sandro> Ivan: A GRDDL transform with do BOTH (2) and (3). 21:48:06 <pfps> +1 21:48:07 <sandro> Ivan: A GRDDL system with do BOTH (2) and (3). 21:48:16 <schneid> ivan: if a grddl system sees several transforms, then it applies them all and merges the resulting RDF documents 21:48:43 <jar> I think DanC is hinting at content negotiation. spec is one 'representation', xslt is another. don't know if that will work. 21:48:50 <bijan> So what's wrong with that? 21:48:56 <schneid> ianh: let's assume the grddl spec is the way ivan and bijan say 21:49:02 <bijan> It can run them both, merge them, and it's fine ;) 21:49:13 <msmith> http://www.w3.org/TR/grddl/#sec_agt , section 7 21:49:16 <pfps> +1 to bijan 21:49:35 <sandro> PROPOSED: if we can have multiple GRDDL transforms, and systems and systems will behave reasonably with them, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); if only one transform, then we'll do the on-line transform service only. 21:50:10 <pfps> +1 21:50:16 <bmotik> +1 21:50:17 <sandro> +1 21:50:18 <ivan> 1 21:50:18 <alanr> +1 21:50:19 <IanH> +1 21:50:20 <baojie> +1 21:50:22 <msmith> +1 21:50:22 <zwu21> 0 21:50:24 <AchilleF> 0 21:50:27 <bijan> In general, that we cannot rely on the spec where it conflicts with the *assumptions* certain people have about the spec. In particular, the people objecting. 21:50:35 <bijan> +1 (for the sake of group hugs) 21:50:40 <schneid> +0.5 (sounds good, at least...) 21:50:43 <sandro> *hugs* bjian 21:50:45 <MarkusK_> +0.5 21:51:25 <alanr> even better alan *hugs* bijan 21:51:27 <sandro> RESOLVED: if we can have multiple GRDDL transforms, and systems will behave reasonably with them, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); if only one transform, then we'll do the on-line transform service only. 21:51:53 <sandro> RESOLVED: if we can have multiple GRDDL transforms, and systems will behave reasonably with them, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); ELSE: we'll do the on-line transform service only. 21:52:10 <sandro> RESOLVED: if we can have multiple GRDDL transforms, and systems will behave reasonably with them, then we'll do (1) the spec, (2) on-line transform service (if a good one is produced), AND (3) download single xslt (if a good one is produced); ELSE: we'll do the on-line transform service only. This closes ISSUE-97. 21:52:32 <bijan> I wonder if we should send a bug report to the GRDDL list 21:54:47 <bijan> q+ 21:54:52 <bijan> zakim, unmute me 21:54:52 <Zakim> bijan was not muted, bijan 21:54:55 <IanH> Q? 21:54:58 <IanH> q? 21:55:02 <IanH> ack bijan 21:56:22 <bijan> http://www.w3.org/2007/OWL/wiki/LC_Responses/JH1 21:56:30 <bijan> Pointer! 21:56:36 <ivan> bijan: can I send it to Jim now? 21:58:25 <alanr> action: Alan to send wg apology to jim re: initial version of http://www.w3.org/2007/OWL/wiki/LC_Responses/JH1 21:58:26 <trackbot> Created ACTION-300 - Send wg apology to jim re: initial version of http://www.w3.org/2007/OWL/wiki/LC_Responses/JH1 [on Alan Ruttenberg - due 2009-03-03]. 21:58:42 <IanH> PROPOSED: Bijan sends response http://www.w3.org/2007/OWL/wiki/LC_Responses/JH1 21:58:45 <pfps> +1 ALU 21:58:51 <IanH> +1 21:58:59 <bmotik> +1 21:59:01 <msmith> +1 21:59:05 <baojie> 0 21:59:06 <alanr> _1 21:59:07 <ivan> +1 21:59:08 <alanr> +1 21:59:09 <bijan> +1 21:59:10 <zwu21> +1 21:59:11 <MarkusK_> +1 21:59:18 <IanH> RESOLVED: Bijan sends response http://www.w3.org/2007/OWL/wiki/LC_Responses/JH1 22:00:10 <ivan> topic: number 8, facet space 22:00:16 <IanH> http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0006.html 22:00:31 <IanH> http://www.w3.org/2007/OWL/wiki/LC_Responses/MS2 22:00:54 <ivan> schneid: there is some text in the struc spec where the topics is facet space of datatype maps 22:01:09 <ivan> ... (reads the content of the text) 22:02:13 <ivan> after looking in the diret semantics a value space of some of the datatypes in the dataype map, for everything else the definitions are not specified 22:02:27 <ivan> bmotik: i have a slight problem 22:02:39 <ivan> ... this makes the definition of one datatype dependent on the others 22:02:51 <ivan> ... you should be able to do them independently 22:03:08 <ivan> ... we do define by taking the definitions from somewhere 22:03:25 <ivan> schneid: we are talking about datatypes 22:03:46 <ivan> ... do we talk about datatypes and these arbtirary objects are in some value space 22:04:03 <ivan> bmotik: it should be possible to define a datatype in isolation 22:04:31 <ivan> schneid: if we do not talk about data values, then can we allow things without a value 22:04:40 <ivan> ... there should be some data value for the facets 22:04:49 <ivan> ... one point what do we want to have 22:04:50 <IanH> Q? 22:04:56 <ivan> ... other what is in the definition 22:05:13 <zwu2> zwu2 has joined #owl 22:05:18 <ivan> ... The way things are defined is that the datatypes are also in the datatype maps 22:05:30 <ivan> ... what you want to have does not match to what is written 22:05:44 <ivan> schneid: i would suggest to have an offline diiscussion 22:05:45 <bijan> zakim, mute me 22:05:45 <Zakim> bijan should now be muted 22:06:04 <ivan> msmith: you primary object is that an arbitrary object can come from another domain? 22:06:19 <ivan> s/diiscussion/discussion/ 22:06:26 <bijan> In rdf, malformed literals get interpreted as an arbitrary element of the domain outside the datatype 22:07:10 <IanH> http://www.w3.org/2007/OWL/wiki/LC_Responses/IH1 22:07:22 <ivan> Topic: number 9 22:07:29 <ivan> bmotik: i have it on my todo list 22:07:45 <ivan> Topic: editorials 22:08:01 <bijan> It was sent 22:08:04 <ivan> IanH: we have a couple here which are done and we should just decide whether we should just ship them 22:08:11 <ivan> .. number 14 done 22:08:27 <ivan> ... number 16: this is response draft, edits identified 22:08:30 <IanH> http://www.w3.org/2007/OWL/wiki/LC_Responses/MS6 22:10:53 <Zakim> +Evan_Wallace 22:11:03 <ewallace> Hi 22:11:10 <IanH> PROPOSED: In response to http://www.w3.org/2007/OWL/wiki/Responses_to_Last_Call_Comments we will use the term "lexical form" for datatypes 22:11:47 <zwu2> +1 22:11:51 <baojie> +1 22:11:55 <schneid> +1 22:12:18 <IanH> +1 22:12:44 <bijan> +1 22:13:23 <MarkusK_> +1 22:14:32 <baojie> http://lists.w3.org/Archives/Public/public-owl-wg/2009Jan/0042.html 22:14:33 <IanH> RESOLVED: In response to http://www.w3.org/2007/OWL/wiki/Responses_to_Last_Call_Comments we will use the term "lexical form" for datatypes 22:14:48 <IanH> http://www.w3.org/2007/OWL/wiki/Responses_to_Last_Call_Comments 22:17:48 <Zakim> -AchilleF 22:24:24 <alanr> http://lists.w3.org/Archives/Public/public-owl-wg/2009Jan/0042.html 22:25:00 <alanr> I don't understand the meaning here - abbreviation is the process of transforming literals of datatype rdf:text isn't it? I assume in OWL that no literals of datatype xs:string would be present as they have a rdf:text form. Or can there be two representations? 22:26:24 <jar> jar has joined #owl 22:31:55 <bmotik_> bmotik_ has joined #owl 22:32:30 <sandro> action: jie Contact Andy Seaborn and try to make sure he's happy with our work on rdf:text, and will talk to use about any remaining issues. 22:32:30 <trackbot> Created ACTION-301 - Contact Andy Seaborn and try to make sure he's happy with our work on rdf:text, and will talk to use about any remaining issues. [on Jie Bao - due 2009-03-03]. 22:32:45 <ewallace> q+ 22:33:14 <IanH> Q? 22:33:20 <IanH> q? 22:33:47 <IanH> ack ewallace 22:33:56 <ivan> zakim, who is here? 22:33:57 <Zakim> On IRC I see jar, zwu2, dlm, AchilleF, schneid, msmith, christine, sandro, baojie, RRSAgent, ivan, MarkusK_, Zakim, bmotik, IanH, bijan, ewallace, trackbot 22:34:12 <bijan> zakim, unmute me 22:34:12 <Zakim> bijan should no longer be muted 22:35:03 <ivan> adjurned 22:35:06 <zwu2> zwu2 has left #owl 22:35:07 <ivan> clap clap clap 22:35:44 <bijan> Ivan has a huge range of facial expressions 22:36:10 <Zakim> -Evan_Wallace 22:36:24 <Zakim> -bijan 22:42:51 <Zakim> -MIT346 22:42:52 <Zakim> SW_OWL(F2F)8:00AM has ended 22:42:54 <Zakim> Attendees were jar, MIT346, Evan_Wallace, Achille, uli, bijan, christine, +1.518.276.aaaa, AchilleF <sandro> Meeting in progress. New content inserted above this line.